Time ‘Series Analysis river enneer on of Trend) m INTRODUCTION Time series has immense importance in the economicand business fields. Most of ty, ana! business data like prices, national income, population, imports, expor's, production, eqns sales, profits, ef, are collected on the busis uf time (such as days, months, years). These a subject to regular and irregular changes constantly with the passage of time. For example, some prices show upw ard trend and sometimes the declining trend, For the measurement of such, chan in the values of the data over time, the study of time series is extremely useful ty @ MEANING OF TIME SERIES-—__ The set of data collected on the basis of time (such as days, wuuths, years) is called a series. In other Words, when datzaré observed onthe bass of ime (days, months or yeas), thes Known as time series. Under time series, there are two types of variables: (i) Independent Vatiah This represents the time, (xt) Dependent variables: This represents changes taking place in te ig of data (population, sales, production, etc.) with the passage of time. The following example illustrate the meaning of time series: z Tablet Tablesll Year Production (in tonnes) Year Population (in crores 1975 50 1931 1976 59 1910 53 1978 55 1979 60 In the above said Table I, production of some commodity is shown between time period 1975t0 1979. Table II contains the population figures of a country in different years. Both of them are the examples of time series. @ DEFINITION OF TIME SERIES Some important definitions of time scrics are as follows: 1. A set of data depending on the time is called time series, —Kenny and Keeping 2. A time series consists of data arranged chronological. —Croxton and Cowden 3. A time scrics is a sct of ubservations taken at Speticied times, usually at equal interv als. —Speigelyee Analysis (Measurement of Trend) a s yTILITY OF TIME SERIES f time serics | ‘ . is re study 0 ics las great world which psn be TOWN pin Importance in economic_and_ business W 1 i gy FoStuly the Past Behaviour ofthe Data: Witne help of time series, changes oceurred in gqastare studied. Only by analyzing the various sorts of changes occured in the past, economists. gd businessmen can rame ther presen policies by taking advantage ofthe past experience, 2) To Forecast Future Behaviour: With the help of time series anticipation of changes going sp ovcucin the future becomes possible because studies about past prove to be very useful fur forecasting about future. — 177 “{G/Estimation of Trade Cycles: Cyclical fluctuations in a time-series.give idea about the charges taking place in the busines like boom, recession, depression and recovery. Businessman can apply this knowledge to rationalise is course of aetion, by way of which potential losses can be, avoided. - — eau With other Time Series: Time series analysis is also important for the coftiparison of ~rious time series By comparing different time series together, their cause and effect can be more elaborately analyzed. ~ (5) Study of Present Variations: Time series a vetlatious in dif is is also helpful in studying the present ne, export-import, price, output, ele. : _ I classes like businessmen, farmers, consumers, economists and goverfiment and accordingly they plan and direct their act 1 COMPONENTS OF TIME SERIES Many ofpesof-changes collectively exer influence on tim components of fime series. A time series h (1) Secular Trend/Trend « 1. (2) Seasonal Variations - § (3) Cyclical Variations - C .~ (4) Irregular Variations - 1 {secur ‘Trend/Trend-T; Secular trend refers to the general tendency of the data to grow or decline over a long period of time. Any time series shows various fluctuations from time to time but inTong period, that series has the increasing or declining trend in one direction. This is secular trend. Secular trend includes or incorporates not short period variations but long-period vari ‘ions. For ‘cxample, if we study the industrial output statistics since 1951, we will find that except for one or ‘wo years the production of industrial goods has been subject to gencral rise. Similatly, prices of the Commodities, money supply, bank deposits. population, etc.. are subject to the upward trend. ~ imilarly, since 1951, there has been persistent decline in death rate per thousand in India. Death ‘ule in some year has risen due to abnormal causes but secular trend is towards decline. Thug, secular ‘rend shows persistent growth or decline in a time series. Quite often, ti ‘end due to population growth, technological reforms, capital formation, impy ‘teanisations, et: series exhibit secular rovements in busicular trend is usually of two types: (a) Linear Trend: When long-term rise or fall in a time serics takes place by ag, amount, then that is called a linear rend. This is also known as straight ling trend Te represented by the following equation: 2 Y=a+dDX (b) Parabolic Trend: The trend is said to be parabolic when long-term rise or fa, series is not taking place at a definite rate. It has many forms but most prominent og the second Degree Parabolic or Quadratic trend. Its equation is as follows: me Yoa+bX+cX? A2) Seasonal Variations-S: Seasonal variations refer to periodic variations in time series , oeCur regularly within a period of 12 months. These variations are of regular type and they ze _AXo be anticipated. Most of the variations taking place in economic and business world are of, type. Prices of foodgrains tend to be low at the time of harvesting and high at sowing time, Dus, winter, woollens clothes rise in price and become cheap during summer. In India, itensilg 3. shoots up on the occasion of Diwali, sale of essential items tends to be high in the first week oft month and tends to be low in last week of the month. Similarly production, consumption, prices commodities, interest rates, etc., move up and down all the year due to seasonal variations. Seas: vatiatiuns are most perceptible during different months or weeks of the year. Seasonal variaice: are affected by the climate and customs. During summer, demand for fans, ice, coca-cola tendstsit greater compared with other seasons. Demand for college text-books tends to be high dais July-August each year. The time period of seasonal variations generally remains definite, (3) Cyclical Variations-C: Cyclical variatianstefer to the oscillatory variations in a timeseis which have a duration anywhere between’2 to 10 years. Cyclical variations arise due to trade qs _“Business cycles have been described ‘as successive waves uf expansion and contractivn tht acenr at about the same time in many economic activitics.”—Burns and Michell. Cycli! variations have four phases (1) Prosperity, (2) Recession, (3) Depression and (4) Recovery. The phases are illustrated by the following yraph: PHASES OF BUSINESS CYCLE Pre ations 25, oe oy is ened Trend Line Depression rs Above graph wakes it clear that business activities (Production, Prices, Employment ie etc.) tend to be at their height during prosperity, (lien recession or the fall starts and gu adually comes the lowest limit of decline in trade activities. This is the stage of depression. This is sina the sequence of cyclical fluctuations but each cycle and its length of phase is different. ACO" ‘y F : Ss an estimate, a trade cycle takes complete turn from 2 to 10 years and then it repeats itself. ™ sequence goes on. . in oh (4) Irregular Variations-1: Irregular variations refer to those short term oman dat series which occur irregularly due to certain accidental causes. These are also know" _—alysis 1(Measurement of Trend) 179 aD ies. wations Inegular varations lake place due o accidental eauses like war. earthquake, floods, rial strikes, etC. They don’t happen in definite order or in a systematic way. Their in’ yement and anticipation becomes very difficult. The decline in industrial output due to the ein story ian example of iregular veratins, ri 1 ANALYSIS OR DECOMPOSITION OF TIME SERIES ‘Time series is composed of four components viz, trend (I), Seasonal variations (S), Cyclical sions (C) and Irregular variations (I). There is always some sort of relationship in these four wn To study the influence of these on time. seri ue comport i Components on time they are measured ately, which is called as Analysis or Decomposition of time series. According to Prof. Speigel: “Theanalysis of time series consists in decomposition of a time series into its basic components.” Models of Analysing Time Series: Analysis of time series is based on two models: (1) Additive model (2) Multiplicative model © (1) Additive Model This model is based on the assumption that tine series isthe sum of the four components. ‘According to «die: formula: O=T+S+C4+I ‘This model treats all the constituents as residuals on the basis of which, hy deducting trend from the original data, short term fluctuations can be determined. Similarly, cyclical variations and ‘regular variations can be determined by deducting seasonal variations from short-term variations. On he basis of additive model, the analysis of various components is illustated ws given below: O-T=S+C+I O-T-S=C+I 0-T-S- © (2) Multiplicative Model This model is based on the assumption that a time series is the product of four components. According to the formula: O=TxSxCxI=TSCI Whatever the component is to be separated, works as a divisor with respect to original data (0), Analysis of different components on the basis of multiplicative model can be expressed in the following forms: TxSxC Us = — “ally, multiplicative model is most often used for time series analysis, bh10. METHODS OF MEASURING TREND ‘in methods of measuring trend in a time series are as follows: The mi (1) Freehand Curve Method (2) Semi-Average Method (3) Moving Average Method (4) Least Square Method Let us study these methods in detail. © (1) Freehand Curve Method This isthe simplest method of trend-fiting, In this method first of all, the original dag time series is plotted on a graph paper. Thereafter, taking care of the fluctuations of data, in curve is drawn which passes through the mid-points of the fluctuations of time series Infy, curve is called as freehand trend curve. This method is also called as trend fitting by inspetie ‘The procedure of this method can he illustrated by the following example: Example 1. Fit a trend line to the following data by the freehand curve method: Year: 1971 | 1972 | 1973 | 1974 | 1975 | 1970 | 1977 | 1978 | 1979] wa] Production (‘000 ¢): | 160 | 185 | 165 | 152 | 174 | 150 | 174 | 175 | 160 | 180 Solution: Fitting of Trend Line by Freehand Curve Method Trend Line Production ('000 %) a 191872 1873.«197A—« 9751876 «SIT «1978 19791860 Year > Merits and Demerits of Freehand Curve Method Merits: (i) This method is simple. - (ii) This method is flexible. (iii) No mathematical formula is used in this method. (iv) This method is also used for forecasting about future. Demerits: | (i) This method is based on subjective judgements. So bias may affect the findings: (ii) There is lack of accuracy in this method. i fefinite. This i* bet (iii) Long-run movement (trend) obtained from this method is not d number of curves can be fitted from the same original data.Yr sine Series Analsi-l (Measurement of Trend) 181 ime EXERCISE 3.1 1. Fita trend line to the following data by freehand curve method: [ voar 1986 | 1087 | 1986 | wey | 1990 | 1991 | 1992 | 1983 | 1994 Salas (‘000 Units): 22 20 24 30 18 26 20 32 6 —— __ @ (2) Semi-Avorage Method In this method, first of alltime series is divided into two equal parts and thereafter, separate arithmetic mean is calculated for each part. The two values of arithmetic means is plotted on graph comresponding to the time periods. Toining the two points, suaight line thus obtained is called as trend line The semi-average method can be applied in case of two situations: (1) When the number of years in a series is even: When the given number of years in a series iseven like 4, 6, 8, etc., then the scries cau be easily divided into two equal parts. In this situation, trend-fiting process can be illustrated with the following example: Example 2. Fita trend line by the method of semi-average to the data given below: Year: 2000 | 2001 | 2002 | 2003 | 2004 | 2005 | 2006 | 2007 Sales (< lakhs): | 412 | 438 | 44a | 454 | 470 | 4a2 | 490 | 500 Solution: Year Sales | Semi-Total ‘Semi-Average Middle Year (@ lakhs) 2000 412 2001 438 _ Pa aaa — 1748 | X= 174845437 | 5 o001.5 2008 454 2004 470 2005 402 _ a + 1942 —2005.5 2007 500182 Trend Line by Semi-Average Method a ‘Actual Line 80 0 Sales (¥ Lakhs) zaeo p01 2002 2003 2008 2005 | 2006 2007 Year Business Se, (2) When the number of years in a series is odd: When the number of years in a series iso like 5, 7, 9, then there wil be a problem in dividing the series into two equal parts. In such cae, te mid-year figure is to be dropped. For example, if 1981 to 1989 (i.., 9 years) figures are given, tes we will delete 1985, .. th year and its corresponding figure and we will make 4-4 years’ pas, i.e., 1981 to 1984 and 1986 to 1989. The remaining process will be the same as before. Trend fting in this case can be illustrated by the following example: Example 3, Fit trend line by the method of semi-averages to the data given below: Solution: Year: 1981_| 1982 | 1983 | 1984 | 1985 1986 | 1967 | Profit (000%): 20 22 ar 26 30 2 | «| Also estimate the profit for the year 1988. average of the two parts will be calculated as given below: Since there are 7 years, the middle year 1984 will be left out and the arithmett 1987 40 Veet Profit Semi-total Semi-average Middle year (009 2) nn Z 1982 2 —+69 | X,=69+3=23 | 1982 1983 a = ee Omitted 1985 30 1986 29 — 99 Xp =99+3=33 7“aa ies AM valysis-1 (Measurement ef Trend) 183 rend Line by Semi-Average Method Graph showing Trend Line o Actual Line ‘Trend Line seer 1621860 10oa dhe Goa dar dow Year The estimated profit for the year 1988 = & 37,000. > Merits and Demerits of Semi-Average Method Merits: (i) This is an easy method. (i) This method is free from bias. (iii) Trend values thus obtained are definite. (iv) Less time and effort is involved in drawing the trend line. Demerits: (@ This method is based on straight line trend assumption which does not always hold true. (i) This method is affected by extreme values. (iii) This method ignores the effect of cyclical fluctuations. Notwithstanding above said demerits, this method is more suitable in comparison with freehand trend method. EXERCISE 3.2 Se | Fita trend line by the method of semi-average to the data given below: [Year: 1981 | 1962 | 1983 | 1984 [1985 | 1086 | i007 26 24 38 40 30 46 [Production (000 tonnes): |__ 30 2. Fita trend line to the following data by the semi-average method: Year: 1983 1984 1985 1986 1987 1988 a2 a5 70 20 95 Profits (‘000 ¥)= 80 3. Fita trend line for the following data by semi-average method: Ye 1980 | 1981 | 1982 | 1983 | 1984 [ 1985 [ 1086 | 1987 LProduction (‘000 units): 12 14 16 20 20 34 28 15 Also estimate the value for the year 1988. Reeist @ (3) Moving Average Method Moving average method is very widely use are calculated. In moving average compureune Year, 4 year, 5 year, 7 year should be tokens Periodicity of data and there is no specitie rule ey © Odd Period Moving Average Gi Even Period Moving Average > @ Odd Period Moving Average When period of the movin; i ig average is odd, s i the computation of moving average: BE ORE SAY 3 years, then following steps ae oe taken, (2) First of all, add up the values correspondit a ing to fi i sum before the middle year (Le., 2nd year) Tt > Yea in the time series and pug (2) Thereafter, leaving the first year value, add up second, third and fourth year Values anda the sum in front of middle year (i.e., 3rd year). Carry this provess further ah We reach last value of the series. G) Moving totals thus obtained are to be divided by the period of the moving averge =i show the trend values of different years. Similarly, five-yearly, seven-yearly moving averages can be obtained. The computation procedure of 3-yearly moving averages can be illustrated with the follovitt example: Example 4. From the following data, calculate trend values using 3-yearly moving average: [vear: 1981 | 1982 | 1983 | 1984 | 1985 | 1986 | 1 Production: 412 438 446 454 470 483_| 400 Solution: Three-yearly Moving Average ie Year | Production Three-yearly moving totals | Three jet inert 1981 412 = poe 1982 438 ——>+ 412+438+446=1296 1983, 446 ——> 438+446+454=1338 1984 454 > 446+454+470=1370 1985 470 > 454+-470+483=1407 1986 483 > 470+483+490=1443 1987 490 od a185 sigs Amalysis-l (Measurement of Trend) 5, From the following data calculate 3-yearly, 5-yearly and 7-yearly moving averages sample vd plot the data on the graph: med 2 Year: 1989 | 1990 | 1991 | 1992 | 1993 | 1994 | 1995 4996 | 1997, Cyclical fluctu: v2[«]o|-2[-1]+2] +1] 0 | 2 ¥ 1998 | 1990 [ 2000 | 2001 | 2002 | 2003 Gyelical fluctuations: | —1 | +2 | +1] 0 | -2 | 1 sotto? Calculation of 3-Yearly, 5-Yearly and 7-Yearly Moving Averages Year Cyclical 3-yearly movin: jarly moving | 7-yearly moving fluctuations averages eS PRBS averages 1989 +2 a = - 1990 +71 + 1.00 ~ ~ 1991 oO 0.33 oO 7 1992 -2 = 1.00 o +0.43 1993 -1 0.33 0 +014 1994 +2 +067 0 = 0.28 1995 +1 + 1.00 o — 0.43 | 1996 oO 0.33 0 + 0.14 {| 1997 -2 = 1.00 o + 0.43 | 1998 “4 0.33 o $0.14 | 1999 +2 +067 0 - 0.28 | ) 2000 “4 +1.00 ° ~ 0.43 ! 2001 oO -0.33 o - | 2002 -2 = 1.00 - - Hi 2003 -1 ba BA _ | = } eee | son yearly 5 yearly | 7 yearly Cyclical Fluctuations ~ Saas Tao Gar Tae eos Won vag abe Two be 185R 2E60 A00H 2002 2005 Years™ Business Say et 186 ie, > (ii) Even Period Moving Average When moving average peirod is even, say 4 years, then moving averages have to be centerey can be computed by two methods: it (a) First Method (6) Second Method. Let us study these methods in detail. (a) First Method: The computation procedure of 4 yearly moving average is as follows: (i) First of all, add up first 4 values corresponding to the first 4 years and put the sum iq between second and third year, Thereafter the next total (i.¢., from 2nd to Sth year total) is tobe put in between 3rd and 4th year. Carry on this process till the last value of the series, (ii) Now add up the 1st and 2nd 4-year totals and put them in front of 3rd year. Similarly, aig up 2nd and 3rd 4-year total and put them in front of 4th year. Carry on this process til the last value. (iii) 8 years’ totals thus obtained are to be divided by 8. These values are 4-yearly moving averages and show the trend values for different years. The computation procedure is made clear by the following example: Example 6. Calculate the trend values using 4-yearly moving average from the following data: Year: 4970 | 1974 | 1972 | 1973 | 1974 | 1975 | 1976 | 1977 | 1978 | 1979 Sales (in crore): 7[/el[el[u][wl]zi{[se]s6{s5| Solution: Computation of 4-Yearly Moving Average (Trend) by Centering the Totals Year | Sales 4-yearly 2 period moving 4-yearly moving (in moving totals totals of 4-yearly | average (centred) crore) moving totals (Trend values) (1) (2) (3) 4) (5) = (4) +8 1970 7 = _ _ 1971 8 - = = > 74949411235 1972 9 — 35+38=73 | 738 =9.125 —— 8+9+11410=38 1973 | 11 —+38+42=80 | 80 +8 = 10.00 — 9#11410412=42 1974 10 — 42441 =83 | 83.48 = 10.375 > 1141041248241 1975 12 —41+96=77 | 778 = 9.628 ——> 10+12+8+6=36_ jes Anas? Measurement of Trend) 187 fa s — 30+ 31267 | 678= 8.375 —> 1280645231 Tel —ot+29=00 | 60-8 = 79 —+ 0+0+5+ 10529 1970] 5 _ _ - 1979 | 10 = _ = () Secoud Method: There is an alternative method of constructing 4-yearly centered moving snerages, the procednre of which is given below: (0 First of all, add up the 4-values corresponding to the first 4 years and put the sum in between second and third year. Thereatter, the next total (i.e., from second to fifth year total) is to he put in between 3rd and 40h year. Carry on this process tll the last value of the series. (i) 4-yearly moving totals thus obtained are divided hy 4 to obtain 4-yearly uncentered moving averages. i) Now add up Ist and 2nd 4-yearly moving averages and divide it by 2. Put this average in front of 3rd year. Similarly, add up 2nd and 3rd 4-yearly moving averages and divide it by 2 and put this average in front of 4th year. Carry on this process till the last value. These values so obtained are 4-yearly centered moving averages and show the trend values for different years. The computation procedure is made clear from the following examples: trample 7. Calculate trend values using 4-yearly moving average from the following data: Year: 1970 | 1974 | 1972 | 1973 J 1974 | 1975 | 1976 | 1977 | 1978 | 1979 Sales (ineroro):| 7 | 8 | 9 | 11 | 10 | 12 [| 8 [6 | 5 | 10 Solution: Computation of 4-yearly Moving Average by Centering the Average Year | Sales | 4-Yearly| 4-Yearly 2Period. | 4-yearly Moving (in crore) | Moving Moving Moving Total Average Totals Average of Col. (4) (Centred) (Not Centered) (Trend Values) (y (2) @) (4) = (3) 24 ©) © 1970 7 = = = _ 1971 8 - a a = _, 875 35 1972 9 — 18.25 9.125 _, 9.50 38 1973 1 — 20.00 10.00 _. 10.50 42We = — Basie Sat wove] 10 — 20.76 10.375 | = 10.28 4 tors | 12 —+ 19.2 9.625 | mar 9.00 36 1976 8 —> 16.75 8375 oats 178 31 197 6 — 18.00 7.500 = 725 29 1978 5 = = - = ss79 |__ tn a “ 4 _ Note: Answers obtained by using Method I and Method II are the same, Example 8. The figures of quarterly income of municipal corporation (in & lakhs) for 2 years given below: Year lst Quarter_| tind Quarter |_ilira Quarter | Ith Quarer | 1995 a 56 48 a 1996 83 52 49 a Using a four-quarterly moving average, estimate the trend values. Solution: Calculating of Four Quarterly Moving Average Years | Quarters | Value | 4-Quarterly| 2Guarterly | @Quartedy Moving totale of Moving Average - totals | 4.duarterly (or Trend Moving totals values) 1995 1 74 56 i aa | 77 | __ gg — 62875 v oa | 7 | __,eog | __, 635 1996 | ag | 7282 —> 505 — 63.125 52 mew — 1a —+ 64.75 mn 49 —— 265 Vv 81 @ Use of Weights in the Calculation of Moving Averages Weights may also be used in the calculation of moving averages. Its objective is to assig® different importance to the values of different years, The following caauples illustrates the procedure of computing weighted moving averages.Solution: Year: 1071_| 197 Values: 2 a 1973 | is74 | 4975 | 1976] 1977 A 1_{ is 3 7 2 ‘omput: : Your 7 tion of 3-yearly Weighted Moving Average wes (¥) | Seyearty Weighted rp ay ee Weighted M.A.* 0) < @),4) ®) (4) 6) 1971 2 “ 1972 8 | xtsexsetetyerr| 1973 4 | xt 4x4 51 6 1974 BL xt set sans A 1975 3 | @xte3c4e7 ed 6 1976 7 | @xterxae2xtyess| 6 1977 2 & * Weighted Moving Average = =" =W Example 10, For the following data, verify thatthe S-yearly weighted moving average with weights, 1, 2, 2,2, 1 respectively is equivalent to 4 years centered moving average: Year: 4989 | 1990 | 1901 | 1002 [1909 | 1994 [1995 | 1990 | 1997 | 1908 Sales (ziniakns)| 2 [6 1]1[5[3]7]2]elalea Calculation of 5-yearly Weighted Moving Average 5-yearly Weighted Moving Ww 5-yearly ee | ae eee Weighted MA. 1989 2 - = 1990 6 a - t991 1 | @xay+2+145)+(1x3)=29] 8 | 29+8=363 1992 5 | (rxeyr2(re5e9+(ta7yeat] 0 | 91-0-s.08 1988 A (txt) #216 43 +7) #(1x2)= 33] 8 33482413 eas 7 | eins) +2Q+7+2)+(1x6)=35] 8 | 35-8248 oy 2 | xaytare2ree(txay=a7| 8 | 378-463 1096 6 | trxr)+2@re+ay(ixe)=39] 8 | 39+8=400 1997 4 _ 1998 8 le 9 Calculate 3 yearly wei following data:ew Business Sri iy Computation of 4-yearly Centered Moving Average Year Sales year moving 2yearly pews a (2) 3) 4 1989 2 = 1990 6 14 — 4991 1 315 5 29 1992 5 16 | out 1993 3) 17 | 5 33 1994 ' 318 Ney 1995 2 19 | As7 wwe | s a2 = 1997 4 1998 8 _ Hence, both the results are the same. © Period of Moving Average While applying moving average method, sometimes the period of moving average is not given in the question. The most important question that arises here 1s about the determination of the period ofthe moving average. Should we take three yearly moving average ot four yearly moving average Troughs the average of these time distances would give the avi should be taken as the period of moving average. The following example would make the procedure clear: Exuemle 31 Determining the petiud of moving average, find trend values by moving averages fot the following data:ries Analysis-I (Measurement oy pe gg Sf Tren) 9 L Annual Sales (in & 000) 42 45 46 52 58 56 51 solution: 72°73 74 18 76 77 18 79 Year {we examine the graph carefully, we have the peaks at the following points Year 1969 |_| 1972 so77|__|4970| | 1982} | 1se5| | 1987 Peak Value: | 42 49 50 45 58 87. 63 Period: a 5 2 3 3 2 Thus, the data shows 6 cycles with varying periods 3, 5, 2, 3, 3 and 2 respectively. ‘The appropriate period of moving average is given by the arithmetic mean of periods of different cycles shown by the data, Hence the period of the moving average is given by: Average Cyclical Period = 34+5+24343+2 _18 6 6Business Sate es Computation of Three Yearly Moving Average [yea | sates | a-vearly] 3-Yearly | Year | Sale | 3-Yearly moving | moving ‘| moving tot averages totals (trend values) 1968 | 40 = = vere | 42 | 140 1969 42 122 122/3 = 40.66 | 1979 48 136 1970 40 126 126/3 = 42.00 | 1980 46 146 1971 44 133 133/3 = 44.33 | 1981 52 156 wiz] 49 139 | 13@/3-40.99] 1902 | 58 166 | 16/0 - 55.33 1973 | 46 137 | 137/3=45.66 | 1983 | 56 165 | 165/3 = 55.09 1974 42 132 132/3 = 44.00 | 1984 51 164 164/3 = 54.66 1975} 44 | 130. | 1303=43.83] res | 97 | 102 1976 44 138 138/3 = 46.00 | 1986 54 174 174/3 = 58.00 1977 50 136 136/3 = 45.33 | 1987 63, _ _ Example 12. Determining the period of moving averages, find the trend values by moving average method for the following data: Year Value _ Year Value Year Value 1 390 6 396 1 459 2 381 7 387 12 438 3 372 8 381 13 435 4 405 9 435 14 492 5 420 10 474 15 510} olution: Since the peaks of the given data occur at the years 1, 5, 10 and 15, the data exhibits regular cyclic movement with period 5. Hence the period of the moving average for determining the trend values is also 5 years, viz., the period of the cyclic variations. Computation of Five Yearly Moving Average Year Value 5-Yearly moving totals Yearly moving averagee (Trend Values) 1 390 — aan 2 381 = = 3 fe 1968 393.6 4 405 1974 394.8yr 193 396.0 397.8 403.8 4146 427.2 4374 448.2 450.6 466.8 @ Measurement of Short-term Fluctuations Time series is the mixture of both trend and short-term flucmations. Therefore, if trend component is eliminated from the original data, we can find out short-term fluctuations. We can use vier additive model or multiplicative model to eliminate trend. ‘The following examples make clear the measurement of short-term fluctuations: sxample 13. Using three yearly moving averages, determine the trend and short-term fluctuations. Plot the original and trend values on the same graph paper: Year: 4980 | 1981 | 1982 | 1983 | 1984 | 1985 1986 | 1987 proft(‘oooe): | te | 21 | 20 | 2 | 2 | 7 | | #2 ution: Calculation of Trend and Short-term Fluctuations using Additive Model Year | Profit ‘000 «) | 3yeary | Syearlymoving/ Short-term (y) moving totale | averages | Fluctuations (Trend (v1) (2) (3) Values-T) (6) “ @) us 1980 18 Lonel ~ - sent a 59 19.667 1.333 ee 5 66 22.00 2 a 25 "4 24.667 0.333 toes _ at 27.000 2 7 ot 30.333 3.333 1985 faa 34.687 0.333 1986 35 _ = 1987 42 = i left is short-term -ainal series (¥), the residual Deducting trend values from co cvven shown in above said column (5). fluctuations. These fluctuations have bee!Business Statig Graph of the O Trend Line \ Profit (000 #) 1950 1981 «1982 «(1983 «19841985 Years Example 14, Eliminate trend by ‘moving average metho Year fst Quarter | lind Quarter | lird Quarter_|_IVth Quarter 2001 40 35 38 40 2002 42 37 39 38 2003, “1 35, 38 2 Solution: Year| Quarter} Given | 4-quartorly | 4-quarterly | 4-cquarterly Given figure Values} moving | moving | moving | percentage of totals | average | average |moving average or centred trend eliminated values A B c D E F G Cx100 SE 2001 I 40 w |] 8 152 38.25 38 i m | 38 ves | 3.57100" 7 155 38,75 0 sede v | 4 20 S100 = 102.56 187 39.25 2002 1 le 425400 _ sog06 39.375 | 39 375 158 305 : 97 x100_ 94 267 hi | az _ = 94,2675yr ps analsisol (Measurement of Trend) 195 series pat ce 156 39.0 m 39 39100. 490.32 36.578) 38.875 155 38.75 38 28 400 = i 3 38.6 38.5 98.70 153 38.25 2003 4 = 41 _ 100 ' a 38.125 |38.125 107.54 152 38.00 0 35 35 100 = 90.9 305 | sg5 156 39.00 uM 38 w_ | a2 Afer eliminating the trend, we are lett with short-term fluctuations. > Merits and Demerits of Moving Average Method Merits: () This method is easy to understand and simple to use. (ii) This method is flexible. i.¢., if number of years are added in a cories, provious calculations are not affected. (ii) This method is most suitable for eliminating cyclical fluctuations. () This method has great practical usefulness. Demerits: (© Ttis difficult to ascertain the proper period of moving averages and if proper period is not ‘ascertained, results will be misleading and inaccurate. (® The second defect of this method is that some beginning years’ and same terminal years’ ttend values remain beyond the scope of calculations. |G The limitations of arithmetic mean affect this method adversely \%) I periodicity in the series is not clearly visible, this method should not be used. a RCISE 3.3 . aadtrena values for the following data, by using 5-yearly moving averages. Also plot the “al data and trend values on a graph: 1969 | 1970 | 1971 | 1972 | 1973 | 1974 [1075 | 1976 | 1977 [1070 672 | 679 | 690 | 702 | 712 | s02 | ao7 | 200 | 16 | ea [Ans. 691.0, 717.0, 742.6, 766.4, 789.2, 811.0] Yey 196, _ - Business S44 2. Calculate trend values hy 4-yearly moving average for the following data: Year: 4961 | 1962 | vous | 1904 | 1206 | 1980 | ree? Do Production ('000): | 13 18 15 24 23 20 24 > (Ans 18, 19.5, 20875 2 3. Assume a 4-yearly cycle, eliminate the trend values from the following data by my, average method. Use additive model ing Year: 1975 | 1976 | 197 | 1978 | 1979 | 1980 | 1981 | 1982 | 1983 | 1984 | 1995, 198g ‘Values: 100 | 105 | 115 | 90 95 85 80 65 75 70 75 a0 (Ans. 101.88, 98.75, 91.88, 84.38, 78.75, 74.38, 71.88.73 13.12, - 8.75, 3.12, 0.62, 1.25, ~ 9.38, 3.12, 239 4. For the following data, verify that the 5-yearly weighted moving average with weight |, 2, 2, 1 respectively is equivalent to 4 years centered moving averages. . Year: 1979 | 1980 | 1981 | 1982 | 1983 | 1984 | 1985 | 1986 | 1987 | 1988 | 1559 Sales (Lakhs): | 5 | 3 | 7 | 6 | 4 [8] 9 | wo] @ | ofa 5. Estimate the trend values using the data given below by taking a four quarterly moving average: YeariQuarter Il i u Vv 2002 78 62 56 m 2003 a4 64 61 82 2004 92 70 63 85 [Ans. 67.5, 68.5, 69.375, 71.375, 73.750, 75.500, 76.50, 77.125] 6. Determining the period of moving averages, find the trend values by moving averages for the following data: Year: 1974 | 1975 | 1976 | 1977 | 1978 | 1979 | 1980 | 1981 | 1982 | 1983 Sales (‘000 Z): 26 29 35 47 51 26 32, 37 46 53 Year: 1984 | 1985 | 1986 | 1987 | 1988 | 1989 | 1990 | 1991 | 1992 | 1993, Sales (‘000 @): 25 30. 36 46. 54 28 31 36. 46_| 54 (Hint: 5-yearly Moving Average) [Ans. 37.6, 37.6, 38.2, 38.6, 38.4, 38.8, 386, 8.2, 38, 38, 38.2, 38.8, 39, 39, 39.39) ee @ (4) Least Square Method This is the best method of trend-fitting in a time series and is most used in practice. This i8# mathematical method and a trend line in this method is fitted or obtained in such a way th following two conditions are fulfilled: (1) 3(Y -Y,)=0, ie., the sum of the deviations of the actual values of Y and cot values ( ¥, )is zero. puted ed_gauyil Mesuremen of Tren) 191 ee ereenasesesitfgeentenennneen One i ; 2 i least, i., the sum sl <) mof the squares ofthe deviations of the actual and computed values from this line is least. thus filled under this methud is called as the Line of Best Fit. method can be used to fit straight line (rend or parabolic trend or exponential tres send line square ad a) Fitting (gine end can be exese by the llowing equation Y=a+bX ‘Trend values, X-Unit of time i of Straight Line Trend whore, Y= se Y-intoroept and bis the slope ofthe line inthe above equation, to determine two constants, a and b, the following two normal equativas esol: ZY =Na+b3X i) LEXY =a DX +b? (ii) After determining the equation Y =a + bX, we find the trend values related to different years splot them on the graph paper whic show a straght line trend. ‘There are two methods of computing straight line trend by using least square method: (Direct Method (2) Short-cut Method (i)Direct Method The procedure to compute straight lin trend in this method is as follows: taken as the year of origin. Usually frst year or before that is taken as zero, (9 Any year is tc. Time deviations are denoted by X: deviations of other years are marked on 1, 2, 3 .. ¢ (ii) Then SX, 2Y ,EXY and ZX7are computed. (i) The values computed are put in the following normal equations: ZY =Na vbEX EXY -aS¥ +hEX* ‘The values of a and b are determined by solving the above said two normal equations. (i) Finally, the calculated values of a and b are put in Y= a + DX and trend values are ___ computed. The following example makes clear the procedure of this method: ‘ample 15, Fita straight line trend by the method of least square (taking 1978 as year of origin) to the following data: Year: 1979 4980 1981 1982 1963 1984 Production (lakh tons): | _9 z 9 10 [12 7 Also obtain the trend values.Business Stain 198 : ned Fitting of Straight Line Tren oe Production | Deviations from 1978 x =a) Yoar au on : 1 4979 5 i ts , 1980 7 7 ” 1981 9 3 3 : 1982 10 4 i : 1983 12 5 : 1984 7 6 102 38 N=6 BY = 60 Ex = 24 zxy=248 | ox2=9 ‘The straight line trend is defined by the equation: Y-atbX ‘Two normal equations are ZY=Na+bEX EXY=aDX +bEX? Substituting the values, we get 60=6a +21b ny 248=21a+91b i) Solving the two equations (i) and (ii) Multiplying (2) by 7 and (ii) by 2 and then subtracting 420= 42a +147b 496 = 42a +1825 By substituting the value of *p? 60= 6a +21B 60 = 6a + 21(2.17) or 6a= 14.43 3 a=2.40 Hence, the trend equation is Y =2.40+ 2.17; origin = 1978, x unit = 1 Year, Computation of Trend Values ; For 1979, x = 1,Y=240+217 (1) =4.57 For 1980, X =2,Y=2.40+2.17 (2)=6.74 in equation (i), we got orpoem Measurement of Trond) ee io For 1981,X~3, ¥=2.40+2.17 @) =g9) For 1982, X=4, ¥=2.40 +2.17 (4) = 11.08" For 1983, X=5, Y 2.40+2.17(5) 13.25 For 1984, X~ 6, Y=2.40+2.17(6) = 1549 goon 16 FHSS aa ia method of east square (aking 1980 as yar of origi) Year: 1980 | 1961 | 1982 | 1983 | 1984 | 1985 | 1986 Production (in ‘O00 units): | 125 | 128 | 433 | 135 | 140 | 141 | 143 Estimate the values for the years 1987 and 1989, Fitting of Straight Line Trend Year Production | Deviatione from 1900 xy cae ™ v0 1980 125, 0 0 0 1981 128 1 128 1 1982 133 2 266 4 4909 195 3 405 9 1984 140 4 560 16 1985, m1 5 705 roy 1986 143. 6 858 36 Ne7 EY=945 ‘EX=21 YXV=2,022 2x2-04 The straight line trend is given by: Y=a+bX li) ‘Two normal equations are ZY=Na+b3X EXY=aZX +bEX? Substituting the values, we get 945=70+21b Ail) 2922=21a+91b Ait) Solving the two equations. Multiplying equation (ii) by 3, we will get 2835 =2la +635 Av) Subtracting (iv) from (ii) we get, 2922-21a+91b 2835 =21a +636 R7= 28b aBusiness Sy Seg By substituzing the value of b in equation (ii), we get From 945=7a+21b 7a =879.753 : a=125.679 Thus, the straight line trend is Y = 125.679 + 3.107X Origin = 1980, X unit = 1 Year. Estimation for 1987 and 1989 For 1987, X=7 Y = 125.679 + 3.107(7) = 125.679 + 21.749 Y = 147.428 For 1989, X=9 Y = 125. 679 + 3.107(9) = 125.679 + 27.963 Y = 153.642 EXERCISE 3.4 f origin) io 1. Fit a straight line trend by the aneth od of least square (taking 1981 as year 0 following data: Year: 1981 1982__| 1983 1984 1088 Value: 15 21 26 33 40 Ans. ¥ = 14:4+6.2X, 144, 20.6, 768, 3.394 Also obtain the trend values. 0 the following data: Fit a straight line trend by the method of least square t —| 19 Year: 1969 1970 I 1971 1972 toe Sales (In lakhs): 45 56 | qe: [tas i 70) ‘Also obtain the trend values. : s os (Take the year 1968 as the working origin) [Ans. Y ee 5 Fit a straight line trend by the method of least square to the following jata: 1984 Year: | 1980_| 1981 1982_| 1983 cs 92 83 94 pe Production {in ‘000 Qtis.): 80 90 rene (ake the year 1980 as the working origin) aa |yr (Measurenient of 201 Anal TO att Method oa Santee inthis method to find straight line trend is as follows fist ofall ae ears ke asthe year forigin value zero and deviations for other a ill be. the deviations wil be eo, Since deviations above tos cae ao iad deviations after middle year will be 1,2, 35-1 1C- an : W middle year will balance out. 1 ; the following example: x lance out, This is made clear by E% gsr eat and 2X are computed. 3) Forcomputing the values ofa, b, we need not have normal equations but they are found by ne following formulae: = 1983_| 1954 | 1955 | 1956 | 1957 7958 _| = = 0 4 +2 8 (a Finally, the calculated values ofa, bare putin the equation Y= a + bX and with is help, trend values are computed. short-cut method is studied in two cases: {@ When number of years is odd. (i) When number of years is even, When Number of Years is Odd ‘When number of years is odd like 5, 7, slusrated with the following examples: Example 17. Fita straight line trend by the method. of least squares to the following data and also er. then the computation of straight line trend can be show on graph paper: Year: 4993 | 1994 | 1995 | 1996 | 1997 | 1998 | 1999 Production (in ‘000 units): | eo | 99 | 9 | 83 | 94 | 99 | 92 Solution: Year _| Production (Y) | Deviations from 1996 (X)_|_(XY) ve) 1903 80 3 ~240 9 1994 90 2 180 4 1995 92 A 92 4 1996 83. 0 0 oO 1997 o4 1 +94 1 1998. 99 2 +198 4 1999 2 3 +216 9 NET LY = 630 X=0 EXY=56 | 5x2 =9R202 Example 18. Fit a straight line The equation of the straight line trend is: Y=a+bX Since, xy =0 aa Thus, ¥,=90+2X Origin = 1996, X unit = One Year. Computation of Trend Values For 1993,X=-3, ¥,=90+2(-3) =84 ¥,=90+2(-2) =86 For 1994, X= For 1995, X: ¥,=90+2(-1) =88 For 1996, ¥,=90+2(0) =90 For 1997, ¥.=90+2(I) =92 Y,=90+2(2) =94 For 1998, For 1999,X=+3, Y¥.=90+2(3) =96 Graph Showing Trend Line ‘0 a Trend Line o " sci Daa ® « “ 2 e| 4 wn ae thod of east aust? iy trend to the following data by the met and (ii) 1997 as origin. Estimate the sa 2002. (i) 1994 as origin, es for1886) 1997 1998 1999 58 78 46 75 @ Fitting of Straight Line Trend by Least Square Method (Origin 1994 = 0) Deviations from xy x? 1994 (x) 1 45 2 112 3 234 4 184 16 5 375 25 N=5 EY = 300 =X =15 EXY = 950 =X? = 55 ‘The equation of the straight line trend is Y=atbx Two normal equations are = 30 950 Na+bEX. LEXY = aEX + bEX2 Substituting the values, we get a+ 15b Sa+55b -@ wii) Solving the two equations. Multiplying equation (i) by 3 and substracting it from equation (ii) 950= 15a + 55b Sa+45b 10b p= Des 10 By substituting the value of ‘b” in the equation (i), we get 300 = 5a + 15 (5) Sa = 225 > a=45 Hence the trend equation is: Y =45 + 5X, Origin = 1994, X unit = 1 year.Busines Stay, (i) Fitting of Straight Line Trend by Least Square Method (Origin 1997 . 6) a 2 Sales Deviations xy x Year fee SN | ome t987 0 1995 45 -2 ~ 90 4 1996 50 eal ~ 56 1 4997 8 0 0 0 4998 46 H 46 1 1999 1B +2 150 4 N=5 EY=300 3X=0 EXY = 50 Wat The equation of the straight line trend is Y=a+bX EY , rar Since, EX=0,a=——, b =A N’ rx Substituting the values, we get a= M6, 4 3 Thus, the straight line trend is Y= 60 + 5X, Origin = 1997, X unit = 1 year. Estimation of Trend Value for 2000 (a) Taking 1994 as orgin For 2000, X= 6, ¥ = 45 + (6) = & 75 lakh, (b) Taking 1997 as origin For 2000, X , Y = 60 + 5(3)= 75 lakh. Note: Its clear that the value of ‘S’ remains the ‘same Whether we use 1994 as origin or 1997 a8 origin but the value uf “a* will differ. Change of origin does not affect the value of ‘b” ‘which is the slope ofthe trend line tit affects the value ut “a" (Hintercept) (i When Number of Years is Even ‘When given number of years is even ‘year becomes a problem. In such a case, (6, 10, 12... et.) in such acase, the selection ofthe middle Corresponding, deviations for other years mean of the two middle years is taken as year of origin and are found out. Deviations will be-2.5,-1.5,-0.5,05,15 and 2.5. To simplify the computation process, these deviations are divided by + or mutipli ‘The remaining steps are the same as before. Example 19. Fit a straight line trend by the method of least square to the following data and estimate the profits for the year 1970, Year: 1961 | 1062 | i303 Profits (& crore): | 80 104 | 1965 | 1966 [ 1967 | 1968. so _|iscz | as [oo [9 | oz | 108yy yo Anatysis-1 (Measurement of Trend) 205 is ee Fitting of Straight Line Trend sol Year Profits | Deviations | ._ Deviations xY x? ” from | Multiplied by 2 1904.5 i 1961 80 35 ie aeaalnatecn 29 1962 90 ~25 aS O55 25 1963 92 15 = 76 9 1964 83 05 a 1 1965 94 405 1 94 1 1968 99 45 3 297 9 1967 92 425 5 460 25 1968 104 43.5 Z 728 49. Nae BY = Fig EX=0 EXY =210 | Ex? = 168 The equation of the straight line trend is: Y=a+bx Since, 2X =0 Substituting the values, we get 734 91.75 8 x = 1.25 (b gives half-yearly increment) The straight line trend is: Y=91.75 4 1.25X; Origin = 1964.5, X unit = : oo Estimation for 1970 For 1970, X= 11, Y = 91.75 + 1.25 (11) = €105.5 lakh Thus, the estimated profits for the year 1970 are % 105.5 lakh. Alternative Method: The same result will be obtained if we don’t multiply the deviations by 2. But in that case the computation will be more lengthy as could be seen below:206 Year Profits Bevistone fom 1964.5 xy ( ri 0 -35 ~ 280 faa 0 -25 225 1963 92 -15 ~ 128 1964 83 -05 -41.5 4968 94 +05 47.0 1968 99 +15 148.5 1967 92 +25 230 toe | 404 3.5 364 N=8 BY = 734 =X=0 EXY = 105 The equation of the straight line trend is: Y=a+bX¥ Since, =X =0 a==F ang » = 2XY N =x? Substituting the values, we get 34.91.75 8 b= = 2.5 ( b gives yearly increment) Uhus, Y = 91.75 + 2.5X, Origin = 1964.5, X unit - 1 ycar, Estimatiun fur 1970 For 1970, X =5.5 Y=91.75 +2.5 (5.5) = & 105.5 lakh, Note:In the second method, the value of b gives annual increment rather than 6 monthly increment as in the first method discussed ubove. It is clear from the example that in the first method, the value of b is half of what we obtained from the second method (b was 1.25 in the first case and 2.50 in the second case). Example 20. Fit a straight line trend by least square method to the data given below. Also find an estimate for the year 2006, What is the annual and monthly increase in sales? 2000 2001 2002 2003 2004 2005) | 28 32 29 35 40 50sy rine slut j. Analysis-1 (Measurement of Trend) 207 rie Fitting of Straight Line Trend Year aan Deviations Deviations xy x? mn wantinted by2 2000 28 -2.5 5 140 25 2001 32 ~15 3 ~96 9 2002 29 05 1 -29 2003 35 +05 4 +35 1 2004 40 +15 43 +120 9 2005 50 +25 +5 +250 25 N=6 | sy=214 EX=0 zxy=140 | sx2-70 The equation of the straight line trend is: Y¥=a+bx Since, ZX =0 a= =F and p = EXY N =X Substituting the values, we get a= 214 W935 67 6 p= 140_, 70 Thus, ¥ =3567 +2X; Origin = 2002.5, X uni Estimation for 2006 For 2006, X=47 Y= 35.67 +2(7) = 49.67 Thus, the estimated sales for 2006 is € 49.67 lakh, Annual increase in sales = 2 xb =2%2=4 Monthly increase in sates =? 2 0.33 lakh Or = Annual Increase 12 0.33 lakh.208 ‘ IMPORTANT TYPICAL EXAMPLES fey, ANT TYPICAL EXAMPLES ' © When there is a gap in the given years Example 21. Below are given the figures of production (in thousand Quinta) op Year: 1ge0 | 1982 | 1903 | 1984 | seas Production: | 77 | 88 | % | 95 | oy 2 Supe, fa (i) Fita straight line trend by the method of least square and Also show, line on the graph paper. ten (i) Eliminate the trend using aditive model. What componens of the ae Ie over? ie so (iii) What is the monthly increase in the production of sugar? Solution: (i) Since there is a gap in the given data. It is not necessary that eviag taken from the middle year would be zero. ions Fitting of Straight Line Trend Year Proton x Ce 1980 7 4 ana 6 | 1982 88 2 176 - 1983, 94 -l 94 1 1984 95 0 0 0 1985 0 4 ot 1 1986 98 42 198 4 1987, 90. 270 @ Ne7 | zy=6%3 DY=-21_ | psy The equation of the straight line treud iy Y=a+6X Sinoe SX isnot equal to zcvo, he values of a and 6 will he found out by solving following two normal equations: 2Y=Na+5DX EXY=aDX+bEX" (i) Substituting the values, we get 633-7a4( 1b Ai a+35b BU Multiplying equation (iv) by 7 and adding (fit) and (iv) 633=7a~5 =14aswrcinent uf Te mi) 1.991 © 2 (appro ) ad athe value of h in equation ii), we Bel rates eteee DO) 5635270 = 3599 7 ps © 7 jon of the straight line trend is soot ° sort Origin = 1984 tion of Trend Values one xen se o0.7l 20-0 =8271 por 1982, xa-2, Ee 9UTIF2-2) 286.71 for 1983, xe, 907426) = 88.71 ror 1984. X= 0, ¥=90.71+2(0) = 90.71 For 1985, ome2(1) 792-71 For 1986, 90.71 +2 (2) =94.71 For 1987, x- 43, ¥90.714203) = 96.71 ‘The straight line can be fitted by plotting trond values on the graph paper Graph Showing Trend Line Trend Line ‘Actual Data Flimination of Trend using Additive Model [Elimination (¥-Ye) | “i a ee 8874 45.29 mf os | an | meTT Bist yy ——_—_—_—_———__ ny, .d, eyclieal and irregular camponciyg or eliminating the tren cour sine Ate ovations will be absent as the data given is annual. weg, 7 ui) Annual increase inthe production of sugar == 2,000 quintas, sy, , ougar= 2 — 22000 | Monthly inerease in the production of sugar= > —"T) = 166.67 quintals, © To Convert Annual Trend Equation to Monthly/uarterlyTrend Equation (oq ve requ trend equation to monthi In certain problems, we require to convert the annual juation ng rercone ao encvnwal trendl equation to monthly trend equation, divide the value of a and the value of ‘6’ by 144, For converting annual trend equation to quater trend equajon 2 the value of a by 4 and the value of b by 16, ig Example 22. Below are given the annual production (in thousand tons) of a fertiliser factor. [Voar: 1977_] 1978 | 1979 | 1980 | 1981 | 1982 [“e 2 Production: 70 75 90 1 95 98. = (a Fita straight line trend by the method of least squares and tabulate the trend yal = nthly and quarterly trend equation, Poy (ii) Convert your annual trend equation into a mo (iii) What is the rate of growth of production per month? Fitting of Straight Line Trend Solution: Year Production | Production xy x) ” a) 1977 70 3 al) 9 1978 i 2 160 4 1979 90 al 90 1 1960 91 oO 0 0 1981 95 4 +95 1 1982 98 +2 +196 4 1983 100 +3 +300 9 N=7 Y= 619 EX=0 EXY = 144 mx’=28 The straight line trend is given by Y=a+bX Since,SX=0, q= =! 619 _ ge 43 N 7 Hl S04 Y=88.43+5.04X; Origin = 1980.For 1977,X=-3, y= 88.43 + 5.04 (-3) For 1978, X=-2, Y= 88.43 4 5.94 2) Bat For 1979, X=~1, Y=243+50401) - es Hor 1980, X=0, Y= 88.43 4 54 o S39 For 1981, X= +1, ¥=88.434 500, seas For 1982, X=42, ¥ 88.434 5.94 a one, For 1983, X= +3, Y= 88.43 4.5.94 @ = 10355 @ ar Quarterly Trend Equation °" 12" 144 88.43. 5.04 eo 2 Yaa Y, =7.369+40.035.X (ii) Rate of growth per month = 5 _5.04 12 ¥.~ 22.1075 + 0.315¥ =0.42 thousand tons, (ToConvert Annual Trend Equation to Monthly/Quarteri Forconverting annual trend equation to'monthly trend equation in case of even number of years, wedivide the value of ‘a’ by 12 and the valne of ‘h? by 6 x 12 (= 72). For converting the annnal trend ‘gation into quarterly trend equation, divide the value of a by 4 and the value uf b by 8. ‘ample 23: Fit straight line trend by the method of least square to the following data : Year: 2000 2001 | 2002 2003 2004 2005 | Profits (& lakh): 10 20 | 20 56 40 ou ly Trend Equation (Even Period) Convert annual trend equation into monthly trend equation. What is the rate of growth of profit per month? Year Profits x xv * (%) 2000 10 -5 =o ee 2001 20 3 ~80 . | 2002 30 ad ay i 2003 56 ca 56 1 2004 40 ae 20 ° | __2005 60 aad ae N=6 EY = 216 =X=0 EXY = 336 =x? = 70aan 212 ZXY _ 336 bat 936 =x? 70 Y=36+48X; Origin =2002,5, X units 4 yea, 3 Year, (i) Monthly Trend Equation will be as: Y.=3+0.066X (Gi) Monthly Increase in Profit =3 @ Shifting the Trend Origin While computing trend values, a certain Year (or unit of time) is Sometimes it may be necessary to shift the origin of the trend equati series. For this, there is no need of making all calculations again bu adjustment to shift the origin of the trend equation: Y~a+b(X£h) Where ‘k’ is the number of time units shifted, If the origin is shifted forward in time Will be Positive and if origin is shifted backward in time, k will be negative, Note: Shifting of origin affects the values of ‘a’ (¥-intervept). The value of 4 ” remains the samy as the slope of the trend line iy the same irrespective of the year uf origin, Example 24, Yon are given the following trend equation: Y = 45 + 5x (Origin : 1990, X unit = 1 year) ‘Shift the origin ta () 1988 (ii) 1993 Solution: y= 454 5X (Origin : 1990) (@ Shifting Origin to 1988: 1988—1999=_9 shift the X by X—2 in the above trend equation, Y =45+5(x-2) —45+5X-10 Y ~35+5x, Origin : 198g (4) Shifting Origin to 1993 : 1993 1990 = +3 shift the origin by 3 years forward. Replace X by X +3 in the above trend ‘equation, assumed as point of ggg ion to some other & t there Will be the folloviy origin by 2 years backwards. Replaceysis] (Measurement of Trend) Anal 7 y =4545 (K+3)=45+5X+15 y = 60+ 5X, Origin : 1993 sales of a particular commodity is given as: pol? . oat 1995 1996 1997 1998 1999 | gales (lakhs ©): 45 56 78 46. 75 straight line trend to the above data assuming 1997 as the year of origin. Betimate 0 Fes he value for 2000. jj How would you shift the year of origin to 1995 in the above problem? Explain. . Hi Convert Your annusil wend equation into monthly trend equations @ ce sot Year Sales (Y) x xy x 1995 48 2 = 90 4 1996 56 a -56 1 1997 78 o o o 1908 46 “1 46 1 1999 75 2 150 4 N=5 EY = 300 3X=0 EXY = 90 2x10 (y) The cquation of the otraight line trend is + bX c Y =60+ 5X, Origin = 1997, X unit = 1 year. Estimation for 2000, X = 3, Y = 60 +5(3) = ™ 75 lakh. ( Shifting Origin to 1995: 1995 — 1997 = — 2, shift the origin by 2 years backwards. Replace X by X~ 2 in above trend equation. | Y =60+5(X-2)=60+5xX— ¥ =5015X, — Origin— 1995; Xunit~1 year, | (ii) Monthly Trend Equation will be as follows: | 7 OU) 0. 2_ x =540.034X 12° 144 ¥ =5 +0.034X214 fs SS att By in EXERCISE 3.5 _— AR ee Odd Period 1. Fitthe straight tine {rend by method of least square to the following data: Yoo 1981 | 1982 [1983 | 1984 Production (+ ): Production (‘000 tons): | 77 88, 94 85 Bae ney aes rcHd Tine to the following data by the least “avers method and timate oy Hy foe it 2000: Prey 1990__| 1992 1904 IO ese Production (000tons):| 16 2 23 2 See ; : 4 What is the monthly increase in production. (Ans. ¥ = 21+0.1.X, Yyops = 21 1 (1000 tons) tho 3. Below are figures of sales (°000) tons: “ny Year: Joo { ose | wus [1008 T1565 | Gps = [ Sales (‘000 tone): | w 14 11 13 * De | Fita straight tine trend and calculate trend values: J fans. ¥ = 13 + 0.75X, 10.75, 11.50, 12.25, 130, ‘4+ The sales of a particular commodity is given as. + Yao00 = 21.6 ('U00 tons), 0.0083, 13.75, 14.59, Wag Lear: 1980 | 1961 | 080 1983_| 1964 | i095 ‘86 ] [ Sates (000 tonsy: 20 2 22 25 26 29 a) (0 Fita straight lin trend wo the above data assuming 1983 asthe Year of origin, Esinge the value far 2000, (@) How would you shift we year of ongin to 1979 in the above problem? Fxplain, (Ams. =25 + 1.642X; Yanoo= 52.914; ¥=18.432 +1 624) 5. Calculate the trend values by the method of least square from the data given helow: = Year: jo92 | 1995 | 1997 [ 1998 | 2000 2o01_| 2003 Value: 75: 67. 6a 65 50, 4 44 (Aus, = 60- 5X, 78, 69, 63, 60, 54 51,5) 6 Calculate the trend values by least square method for the following time series: [Year 190 { 9st { 1982 | 1983 | 1904 | 1965 | 1086 | 1007 | some [Satee (0007 €), 23 | 79 | 76 | 66 | 69 Also calculate the trend values (Ams. (9) ¥ = 7867 +4, a7 {| 70 | 06 | 10 by taking 4-years moving average petiud, 60.07, 64.72, 69.37. 74.09, 78.67, 93, 32, 87.97, 92.62, 97.27 () --70.50, 73.50, 74.875, 78.875, 86.875,"urement of Tr \ ies? | Pst data give the value of sales of a company for the years 1988 t0 1998: 1968 | 1989 | 1990 | 1991 | 1902 | 1990 | 1994 | 1995 | 1996 | 1997 1998 3a vos tak £) (0 500 | 38.5 | 43.0 | 44.5 | 389 | 36.1 | 32.6 | 90.7 | 41.7 ad | 338 s ove. al fe the method of least square to fit a straight line trend to the data given aby ‘a unit @ eSmpute the trend values for 1991 and 1996 (Take X =0 for the year 1993 and the orX as | yeat) W Construct a five year moving average and compare the trend values for the year 1991 and 1996. [Ans. (i) Lincar trond equation ; Y — 39.9 ~ 0.767%; x= X- 1993 Trend values for 1991 and 1996 are 41.4 (lacs &) and 37.6 (lacs &) respectively- (ii) S-yearly M.A. values for 1990 to 1996 are (in lakhs 2): 42.6, 40.2, 39.4, 38.6, 38.0, 38.4, 37.6] method of least square to fit a straight line trend for the following data. (Take 1986 as ysethe ” origin). Year: 1984 1985 1986 1987 1988 production (in ‘000 Qtls. 28 38 46 40 36 ‘Also compute the trend values. What are the annual and monthly increase in production? [Ans. Y= 41.6 + 5.8X; 30, 35.8, 41.6, 47.4, 53.2; Annual Increase = 5.8 thousands Quls. Monthly Increase = 0,483 thousand Qtls.] yen Period Fita straight line trend by the method of least square. Also estimate the production for the 9 eat 1995. What is the monthly increase in production? Year: 1989 1990 1991 1992 1993 1994 Production (in lakh tonnes): 25 40 47 59 68 80 [Ans. Y = 53.17 + 5.3X, 90.27, 0.883] 10. Fitalinea Wend to the following time serics by the method of least square and also obtain the trend values: Ye 1954 | 1955 | 156 | tor | 1958 | 1959 Production (in crore of &): | __7 10 12 14 7 24 [Ans. Y = 14+1.54X; 6.3, 9.38, 12.46, 15.54, 18.62, 21.7] 11, Fita straight line trend by the method of least square to the following data: Year: 1988 1989 1990 1991 1992 1993 Sales (in lakhs): 12 18 26 35 29 42 Estimate the sales for the year 1995. [Ans. Y = 27 + 2.74X: ¥}995 = 51.66] a6 = Cishts ofa calf taken at weekly interval nl calculate the average metiioal of feast square a | 4 5 65.0 | 70.2 | 75.4 | 61.1 | o7.2 Also eatimate the weight of the calf when age is (7) 12 (ii) 15 weeks. [Ans, (i) ¥ = 79.62 + 3.08X, (ii) 2%3.08 = 6, Gap in the Given Years 13, Below are given figures of production (in thousand tons) of a sugar factory Year: tose | 1988 | 1989 | 1990 | 1991 | 1909 Le) Production: 77 88 94 85 of 98 a ()) Fita straight line trend by the least square method and tabulate the trend values (ii) Eliminate the trend, What components of the time series are thus let over? (iii) What is the monthly increase in the production of sugar? [Ans, Y=88,803+1.38X; 83.28, 86.04, 87.42, 88.80, 90.18, 91.56, 957 Monthly increase = 0.115 thousand FH} ng time series shows the costhunit (&) of a product for the period 1981-195 14, The followi 1981 | 1982 | 1964[ 1985 | 1987 | 1090 | 1991 | 1992 | 1009 | 1095 Years: Cost/Unit (2): 332 [ 317 | 357 | 392 | 402 | 405 | 410 | 427 | 405 480 | Calculate the trend values by the least square method and estimate the cost per unt for 1997 [Ans. Y= 388.5 + 7.509X; Yip = 45608) .rsion of Annual Trend Fquation to Monthly/Quarterly Trend Equation Conve 15, The following data relate to sales of Reliance Lid.: Year: 2000 2001 2002 2uus 2008 Sales (¢ lakhs): 40 80 120 200 160 (i) Fit a straight line trend by the method of least syuare and tabulate the trend values (ii) Bliminate the trend using additive model. What compoucuts of the time series are ths left over? (ii) Estimate the likely sales for the year 2006. (0) What is half yearly, quarterly aud wionthly increase in the sales? (0) Convert the trend equation into (a) on monthly basis, (b) on quarterly basis. (vt) Shift the origin (a) to 2004, (b) to 2000. [Ans. (i) Y = 120+ 36X; (ii) only cyclical and irregular variations, (ill) Yanos = 264, (iv) 18, 9, 3 (v) (a) Y= 10+ 0.25X, (8) Y= 3042. 25X (vi) (a) Y = 192 + 36X, (6) Y= 48 + 36%)grement of Sho i Mea otlowing data relate to sales of Bharat Lig. 16 2000 (ii) Eliminate the trend using left over? (aii) Estimate the likely sales for th (iv) What is annual increase in the (9) What is monthly increase in ean ee Bee exPected sales would have equalled to its target of 84 lakhs. (Ans, additive ‘© model. What Components of the time series are thus 'e year 2006. sales? the sales? OY =36 +4.8x (ii _ 2-16,~ 1.2, 15.2,- 10.4, 0, C & Tare left over: on (i) 69.6 (iv) 4.8% 2=9.6 (v) 0.8 lakh : . ~) Target is expected to be attained in the year 2007.5 (ie., 2002.5 + 10/2)] 17. Fitastraight line trend to the following data by the method of least ‘square after summing the quarterly data to yearly data: Year Qa @ a % 1993 10 13 14 12 1994 12 14 15 43 1995 13 15 18 14 1996 15 18 21 18 1997 15 22 23 20 Also find out short-term fluc [Hint: See Example 36] (Ans. Y= 63+8X; Trend: 47. 55, 63, 71, 79; Short torm fluctuations: 2,-1,-3,1, 1) tuations for the given years using additive model, ifting of Origin 8. Production of a particular commodity is given as: : Year: 1997 1998 i999 _| 2000 | 2001 2002 2003, Production of Steel (in tonnes): 60 72 75 65 80 85 95 @) Fita straight line trend to the above data assuming 2000 as the year of origin. Estimate the value for 2004. ; , i) How would you shift the year of origin to 1997 in the above problem? Explain, Wii) Coy \d equation to monthly trend equation, cs — (Ans. (i) Y = 76 + 4.857x, Y 2004 = 95. 428, (ii) ¥ = 61.429 + 4.857X: Origin 1997; (iii) Y = 6.33 + 0.0337X]Ns me : wi [ABC Co. Ltd. is described by the folly; Beata 19. The trend of the annual sales ae Cee 30 + 3.6X, Origin = 2001, X un" ear. innual sales, Convert the annual trend equation into monthly tend equation, Taney 20. Given the following trend te Y, = 84.26 + 5.8 X (origin = 1° 185 and (ii), 1969 ; '5.8X, Origin : 1985 (ii) Y = 32.06 + 58 y 8X. Oty igin to (i) 19% Shift the origin Os 28 a tion is: X unit = 1 year. [Ans. Y= 44,05, company is described by the following equation; $+, mn: X unit = 1 year) 21, The annual trend equal y =50+2X, Origin: 2000, Shift the origin t0 1997. The trend ot annual sales of act y=15+05X ; (origin: 1987, X unit = 1 year, Y unit = Convert the equation to quarterly trend equ > GB) Fitting of Second Degree Parabol rend or Quadratic Trend There may be many such situations in economic and Business elds in Which a staigh trond any at represent the long-term tendency of the time series data, In sul cases tlie NEEM ie fitted. The equativu of the second degree pavabolctey degree parabolic trend or quadrat or the quadratic trend is: Y=a+bX +cX? Where, ais the Y-intercept, 2 is the slope of the curve at the origin and c is the rate of changeia ny 22. Annual Production) ation. [Ans, slope. ne Sad ee ee suas, the values of the constants a, & and c are obtained by solving BY =Na+bEX +cZX’ SEXY - aX +bEX? + cDX? sn EX?¥=aEX? +bEX3 +cEX* as nee Meso! Ite dao akon fom the middle year (oe reine of equations arc reduced to te following: 2x would be zero (ie. EX =0 and EX? =0) and the ZY =Na+clX? EXY =bEX? ae When further solved, ee X? +cEX4 o | these equations gives the following values of th Be constants a, b and ¢. From equation (ii) we get h = 2X¥ 2x”ralysis-l (Measurement of Trend) 219 eS Tren) ——_ a N 7 from equation (We get a= * Jen N-EK4 (py)? “After finding the values ofa, b and cin the above manner, the trend equation can be fitted to otsin teed values ofthe given time series by Substituting the respective values of X therein. Stor-cut method is widely used for fing second degree parabolic trend. and rom equation (i), we yet e= Inpractice, Note: procedure ‘The computation of second degree Parabolic ‘tend/quadratic trend involves the following steps: (i) Find te time deviation of cach year from the middle year and denote it by 'X". (i) Then EX,ZY,EXY,EX7Y,EX?, DY Sand EY 4 ae computed. (iii) The values computed are Put in the above formulae of b, aandc, (ov) Finally, he calculated Values of a,b and e are Putin ¥ =a +DX +cX7and trend values are computed, @ Type I: Odd Number of Years Example 26. Fit a second degree parabolic trend (Y =a+ bY + eX) to the following data: Year: 1909 1984 1985 [ 1986 1987 Production (in crore ¢): 5 7 4 | 8 12 Also compute the trend values, Predict the value for 1988. Solution: Fitting of Second Degree Parabulic Trend Year | Production | Deviations | XY. xy ye ot xt (Y) | from 1985 ®) 1983 5 2 -10 20 4 38 16 1984 7 -1 7 7 1 - 1 1985 4 oO 0 0 0 0 0 1986 9 4 9 9 1 + 1 1987 12 +2 a4 16 NES rY =37 EX=0 | EXY=16 Ext 234 ‘The second degree parabolic trend is given by the equation Y=a+DX +cx? ‘The three normal equations are: EY=Na+bEX +cZX?rx] yyyed Substituting the va i7= 162a(0) +H 42a(10) rom (ii) equation: 16= Multiplying ( bY 2a 0= py 4eyx? +6 Hues, WE ad subtracting it from 84= a= 24 DEN? +c3X" get 450) rel) (10) +e(0)=7 16= 10 y+ (0) + C34) = 84 = 1004 Me 4 16 37=5at 10c 10b > b= 55 ) 10a +34¢ 10a+20e ide 10_ o7t Putting the value of 31= > 5a=29 ‘Thus, the sevond degree P y=64 16x +0.71X7, Ce ‘omputatior For 1983, For 1984, For 1985, Fur 1986, X =+h For 1987, x-+2, Predicted value for 1988 For 1988, X=+3. © Type ll: Even Number of Years x X=-L X=0, Example 27. The price of a commodity during 1 parahala tothe following data. Calculate the tre ‘commodity in 2001. yn of Trend Values 2, 14 cin (i), we get 5a+ 10(0.71) 9 = 5.98 6.0 warabolic trend is Origin = 1985, Xunit= lye 1(-2)°= 5.64 Sl 6.00 = 831 ¥-=6 +1.6(-2)+0-7 ya6 +1.6(-D +0-7-D? = ¥ =6 +1.6(0) + 0.71(0)" y=6 +1.6() +0.71()" ¥=6 +1.6(2) +40.71(2)? = 12.04 is given by: ¥ =6 +1.6(3) +0. 713)" = 17.19 yndd degree slow. Fit a seco! of the 1d estimate the price 1998 1908 994-99 are given bel nd values ani Year: | 1994 1997, 1996 1995 181 192 107 128 wT Price: 100(Measurement of Trend) anatysis-1 (Measure , eset eviations | Deviations wer’ Pyaar [Pace | Devito ‘mated 5 1996.5, by 2) = %) 25 \ 625 100) 2.5) 8 ~800 | 2500) eI aay Bt oe 107 15 3 ~321 | 963 es - 4 toe | 128 08 | 428 | 20 1 wh 5 ae 181 +15 +3 +543 | 1620 ° 425 \ 625 oo] ime | 18 4 Ser ees \ = xX =0 Ay \y x2 =70l2x wn" yaaa The second degree parabolic tend 8 given by the cyuation Y=a+bx 4x? The three normal equations are: ZY =Na+pzx +cEx? EXY = aBXx + ppx? +cEx3 2X?Y = yx? + DEX? 4 oxx4 Substituting the values, we get 848 = 60460) +706 = 848 = 604706 oA £94 = (0) +570) 4 o(0) 694=70b 10160 =70a + b(0) 4 M4M4e= 10160=7004 Mi4e From equation (ii), 694 = 70» Sb= we 9.914 Multiplying (® by 79 and Gii) by 6 ana Subtracting (i) from (iz iplying ig 60960=420a .g4gqe 59360 = 420.4 +4909 « | 1600 = 3584¢ © = 00 446 3584 Putting the value of cin (i), we get 848 = 604 710 (0.4464) =, 848 = 604. 31.248 . a= 816.752 ~~ = 136.125a 2 be, ‘Thus. the second deere’ parabola is oe g14X +0. 4464X°, Origin : 1996.5, * Xuniigt | ja EXERCISE a +bX +cX7to the following: dat 1986, I 1907 5 yo 136.125+9° i id Values Computation of Tren = .914(-9 = For 194, X= 7% y2136.1254 9.914(-9) + 0.4464( 53? sons , ¥=136 125+9.914(-3) +04464(-3)? = 119 1) + 0.4464(—1)? Z 14006 (D (-) i26aye 2136 .125+ 9-914 25 +9.914(1) + 04464(1)? = 146.4854 15 +9.914(3) +0.4464(3)? = 169.884 yn of xe Y xeth ¥=136.1 = 136.12 For 1996, For 1997. 3, Kat Y 136.1254 9.914(5) + 04464(5)* = 196.855 For 199! For 1999, X=+5 ¥ Predicted value for 2001: For 2001, y =49, Pa 136.125+9.914(0) + 0446409)" = 261.5094 for the year 2001 is 261.5094. ‘Thus, the likely price af the commodity 988 | 1090 ] nf aw] [es | 1. Fita parabolic trend ¥ Year: “get | 1982 | 1989 | 1984 | 1985 7fo {7 ele Output (in *000 units): ‘Also compute the trend values. Predict the value for 1990. (Take the year 1985 as working origin) [Ans. ¥ =10.02+ 0.85% -0. 27X?, Trend values: 2.3, 5.04, 7.24, 88, 10.02, 10.60, 10.64, 10.14, 9.1, Yo= 752 a second degree parabola ce of the commodity in The prices ofa commodity during 1981-1986 are given below. Fit a ollowing data. Calculate the trend values and estimate the Pris 1987. 1985 706] 1983 1984 218 120 138 152 ", Trend values: 114.40, 109.12, Yigg = 266321 [ 101 | 1082 a0 | 114 Price: [Ans, ¥ =124.154+9.6X +1.53X7 116.08, 135.28, 166.72, 210.40, *. a second degree parabola (Y =a +bX +cX”) for the following data: x 4 2 | 3 4 5 39 46 [= 5 23 | 33 [Hlint: See Example 40] [Ans, 7=22.78 + 146x+ 0.648 1 Year:223 © 10 the data Biven below and estimate the value for : 1993 190. ‘Year: 4 fa gates (in ‘000 &): no) 12 [5] 1996 1997 1998 10 8 "4 (Ans. y=) ‘i ; -Y =11.031- sates of a particular commodity is given as: 1-0.143X —0.3125X?, ¥jgyy= 1995.5] § 1981 | ogo yee 1983 | 1984 ce (900 tons): 13 13 a a t9a5_| 1986 | 1987 54 60 83 (p Fita second degree parabola (Y = a45x 4 ox? the year of origin and estimate the value for 198 (ji) How would shift the year of origin to 1980 [Ans. (i) ¥ =30410% 4 to the above data assuming 1984 as in the above problem? Explain. ox? » Origin: 1984; 110 (a) Y =14—4X +2X7] «, Caleulate trend values forthe following data using second degree equation: Yoar: 1984 1985 | 1986 | 1987 | 1988 | 1980 | 1990 Output (000 tons): | 100 107_| 126 | 7440 “Ger ee__1880 ‘Alsu make a forecast for 1993. (Ans. ¥ =149.10+18.68X +0.16X?, Trend values: 94.49, 112.38, 130.58, 149.10, 167.94, 187.10, 206.58, Forecasted value 66.94] a ef tS ur > (0) Fitting of Exponential Trend If the time series is increasing or decreasing by a constant percentage rather than constant absolute amount, the fitting of exponential trend is considered appropriate. Such tendency is found iamany economic and business data. The equation of the exponential trend is Y =ab* where a is Y-intercept and b the slope of the curve at the origin of X. In the logarithmic form, the above equation is written as under: logY =loga +X logb When plotted on a semi-logarithmic graph, the curve gives a straight line (or called logarithmic Sinight line). However, on an arithmatic scale chart, the curve gives a non-linear trend. Under the method of least square. the values of the constants a and b are obtained by solving the ‘slowing two normal equations: ElogY =N -loga+log bUX 2 ¥(X log ¥) = log aX + log bEX224 ee Bs Hac, Short-cut Method: Ifthe time deviations are taken from the middle year (oy ath, ), the value of EX would be zero (i.¢..EX =0) and the normal equa te ‘Wations apg two middle y (0 the following: y YlogY-N loga or a > «~All N Xo ; DU Iogy) =log 6X? or 1ogo=2E aD = b= Antilog| logy EX Bx? i After obtaining the valucs of «and b in the above manner and substituting thr yap exquation, we can fit the trend equation under this method and with such an equation yen the trend values of the time series and predict the values for the future year, ob, Note: The annual at of growth in case of exponential trends obtained as =(Antlog 1, tn > Procedure The computation of exponential trend involves the following steps: (0 Find Ure time deviation of cach year from the middle year and deuute it by *Y° (i Obtain the logarithms of the variable Y. (iii) Multiply log ¥ by the corresponding time deviation ‘X” and obtain X logy, (iv) Square up the time deviation X and obtain EX ?, (») The values computed are put in the above formulae of log « aud log b. (vi) Finally in calculated values of log a and log b ate put in log ¥ = log. a+X logb, (vii) Take the antilogs of these logs to arrive at the actual trend values. @ Type I: Odd Number of Years Example 28, Fit an exponential trend Y = ab* to the following data: Year: 1990 1991 1992 1993 1994 1905 | 1006 12 10 14 18 20 24 30 Sales (< crore): Estimate the sales fox dhe year 1997. Solution: Fitting of Exponential Trend Year Sales x log ¥ Xlog ¥ xe (4 1990 1 -3 1.0792 = 3.2376 8 1991 10 2 1.0000 -2.0000 4 1992 4 1 1.1464 = 1.1464 1 1993, 18 0 1.2553 0 z 1904 20 "1 1.3010 1.3010 i 1995 24 2 1.3802 2.7604 . 1996 30 +9 1.4771 4.4313 2 bi BX=0 Zlog¥ EXlogY rx? =28 =8.6389 =2.1090js-1 (Measurement of Trend) | 225 ‘The equation of the exponentiay rend is — Ysab* ‘aking }gArithMs OF both sides, we faye “ff log Y=10g.0+% logy sine sx =0 eli) Zlogy _ 8.6389 7 SLL log b= 2X logY _ 2.1099 EX? gg =0.075 Thus, the equation of the exponential trend in logarithmic form is: log Y= 1.2341 +.0.075x or Y.= Antilog [1.2341 + 0.075%] Here, origin = 1993, X unit = 1 year. : kstimation for 1997 For 1997, X-44, log a= Y= Antilog [1.2341 + 0.075x) >he Antilog (1.5341) = 34.20 crore Thus, the estimated sales for 1997 is 34.20 crore. 49. The sales of a company (in lakhs of &) for the seven years are given below: Year: 1990 1991 4992 1993 4994 1995 1996 Sales: 32 47 65 88 192 190 278 Find out the trend values hy using the equation ¥ = ab* and annual rate of growth. sauion: Fitting of Exponential Trend Year Sales x log ¥ Xlog Y x? m 1990 32 -3 1.5052 4.5153 9 1991 47 2 1.6721 - 3.3442 4 1992 65 -1 1.8129 = 1.8129 : 1993 88 o 1.9445 o 0 1994 132 +1 2.1206 2.1206, 1 1995 190 42 2.2788 4.5876 4 1996 20 +3 2.4393 7.9179 a mo | ae, | The equation of the exponential trend is Y=ab* In the logarithms form, the equation is written as log Y=loga +X logbmo Bee Type Il: Even Number of Years ution: 3. 3 nas = 1.9676 43257 ©9154 28 b= Aniilog [0.154] = 1.4256 Since, ZX =0 fa elon jog a=, EX logY log 6= 3 1B mx? > log Y = 1.9676 +0,154X Computatio For 1990, X For 1991, X= For 1992, For 1993, X For 1994, X=+h Y. For 1 Kor 1996, X=+3, ¥, Computation of Annual Growth Rate n of Trend Values ¥, = Antilug[1.9676 +0.154(~3)]= Antilog( S056}= 94 903 7, = Antilog [1.9676 + 0.154(-2)] = Antiog 6596 2 = Antilog [1.9676 + 0.154(—D] = Antilog[1.8136} Y, = Antilog [1.9676 + 0.154(0)] = Antilog 1.9676 = Antilog [1.9676 + 0.154(1)] = Antilog [2.1216] 995, X 42, ¥, =Antilogl1.9676 + 0.154(2)] - Antilog{2.2756] = 1884) = Antilog [1.9676 + 0.154(3)] = Antilog [2.4296] = 268.1 =(Antilogh 1) x 100 /Antilog (0.154) — 1] x 100 =(1.4256 ~ 1) x 100 = 0.4256 x 100 = 42.56% = 42.56% jon in logarithmic form is: or Thus, the exponential trend equa Origin: 1993; Xunit=1 year (b-1) x1 = (1.4256 - 1) x 100 Ni 00 ample 30. Fit an exponential trend (Y = ab%) to the following data: Year: [teat [1951 [1961] 1974 | 1981 | wet Population ofindia (in erore): | 219 | 36.1 | 439 | 548 | 685 ae Also predict the population for 2001. Fitting of Exponential Trend Year | Population(y) | x ly ¥ Xlog Y x 1941 31.9 -5 1.5038 = 7.5190 % 1961 36.1 -3 1.5575 ~ 4.6725 9 ae 43.9 -1 1.6425 = 1.6425 1 a 54.8 + 1.7388 1.7388 1 ai 933 +3 1.8344 5.5032 8 nee Es +5 1.9263 9.6315 2% 3%=0 | Slogy =10.2099 | =X iogy =3.0996 | »x?=7 099 |eme Sores Analysis-I (Measurement of Trend) 27 The equation of the exponential trend is y=ap* In the logarithmic form, the equation is written as log Y=loga+X lop b Since, EX =0 log a= = log N log b= 2X18 _3.095 5 xx? 70 ‘Thus, the exponential trend equation in the logarithmic form i log Y=1.70 +0.043X, Origin: 1966; X unit=5 years. or Y= Antilog {1.70 +0.043X] Further loga=1.70 = a Antilog [1.70] = 50.12 log b=0.043 => b= Antilog [0.043] = 1.10 ‘Thus, the equation of the exponential trend is Y= 50.12(1.10)* Prediction for 2001 For 2001, X=+/, Y = Antilog {1.70 +0,043(7)] = Antilog [2.001] = 100.2 crore. Example 31, Fit a logarithmic straight line tothe following data: Year: 1901 1992 41003 1004 1905 1996 Production (000 tonnes): | 64 [ 70 | 75 | 2 | a8 | 95 Solution: Fitting of Logarithmic Stralght Line Year | Production] x log ¥ Xtog Y x? ” 1997 i 5 1.8062 9.031 25 1992 70 3 1.8451 5.5353, 9 1993 75 a 1.8751 1.8754 1 1994 82 at 1.9138 1.9138 1 1995 88 3 1.9445 5.6995 9 1996 9% 5 sgrrt 9.8885 25 Ne6 EX =0 | Blog =11.9624 | EX logy =1.1944 | 5x2 79 The equation of the exponential trend is Y=ab*BB ty Inthe logarithmic form, the equation is written as a, log V=loga 1 X log ULoEatitn 0 "hy, Slog _ 11.3624 t, ean 5 = 1,893 log a= 6 7 rx logY 11944 = 9 4 ex? 70 Thus, the equation of the logarithmic straight line is; log Y= 1.8937 + 0.017%; Origin: 1993.5; x upg 1 ze ot ¥ = Antilog (1.8937 + 0.017X] 2 > Mcrits and Demerits of Least Square Mcthod Merits: (0) This method is far better than moving average method because the tend valyy fra yeas ae obtained, Nut even a single initial or terminal trend values 1 left inethod. (if) Ieresults in a mathematical equation which may be used tor forecasting, (iii) It is widely used method of fitting a curve to the given data. The results tnd reliable and appropriate. Demerits: / (9 The computation process in ths method is complex which is not easily undersindy. (i This method does not have the attribute of flexibility. If eome figu subwracted from the original data, all computations have to be redone, (if) Ieis difficult to select an appropriate type of equation in this method, Results based inappropriate selection of equation are likely to be misleading. omit HES ATE aed iy iXERCISE 3.7 SSS Fit an exponential trend ¥ —ab* ty the following data and calculate the trend values. Alo estimate the trend for 1992, Year: | 1985 1986 1987 1988 1988 Sales (€ crore): | 100 15 112 120 130 (Ans. log ¥ =2.0527 +0.0286Xor ¥ ~112.9041.07)", Trend values: 98.97, 105.71, 112 90, 120.59, 128.80 Sales for 1992 = 156.93 crt]sis-I (Measurement of 7 gp aan Teng) 3 i tion figure = peo ion fig SOF Inka ae ive = : wus Year! 1911 Lee Lepulaion cm crorey: | 25.9 = ‘pop! Ba | 1951 [ager [1971 xponential trend Y= abX iy ys S18 | 301 [aso [sar jane : ; es. Estimate t above datg ent yalues. he population ing data by the Method of " : ne fllowing exponential tong 81 Tan of ast square and ind the the folloy rT ns. ¥=33,60 (1.142) 4. ane origin to 2005, SAUBtOn (origin 2001, Yayge tet a p = » X-unit= | year, ¥ - annual profits) 15(17)* year, Pp consumption of clectricin (Ans. ¥ =12.598 (1.1 aa fed as under: ' in the agriculture Sector during the aes = roootd peri 1999 was Year: 1991 | 1992 | 1995 ‘Consumption 50 | 65 (000 lakh tonnes): Find: (i) The second degree (ii) The exponential trey 1 284 | 1095 | 1996 | 197 | 1998 | 1999 5 65 90 95 Polynomial equation, nd equation, [Ans () ¥,~ BS. Y= 714-+3.9% + (-U30)K2 (i) ¥,= (0013) (1.053) ee sCELLANEOUS SOLVED EXAMPLES }. Calculate trend i ple 32. Values from th following data by using four yearly moving average: Neat: Vaue | Year Tae ie a 1973 60 1973 6t a © 1974 85 1981 7 1975 48 1982 8 1976 53 4983 % 1977 or 1984 84 1978 " on; Tour ycarly moving averages can be obtained as follows: Year y Ayearly 2 period 4-yearly moving totals | movingtotal | __ moving ofcols.(3) | _ average (Trend valuos) q (2) @) (4) (5)=(4)=8 1972 4 = - = 1973 61 — ae = ord 55 205 7 52.75Example 33. Solution: B00 Fa “ir 440 ve 53 223 453 ‘wr 67 230 472 1978 62 242 496 ao 60 256 518 a 67 262 540 ra 73 270 ore 1982 78 294 60s 1983 76 att ae - 1984 84 — = ~ II data by —™ Determine the trend for the following () moving average of length 4, and (i#) the least square method. Quarters. > Year t uw im | 1991 210 191 216 200 1992 218 197 230 an 1993 246 215 235 225 @ Moving Average of Length 4 Year Quarter | Given Figures | Four-ngure | 1wo-rigure | Four-figure moving | moving | — moving total total Average 1991 I 210 u 191 B17 Ww 216 95 1642 1642 _ a5. 25 ates Vv 200 831 1656 207.00 1992 1 218 845 1676 209.50 ul 197 866 1701 212.625 ul 230 684 1740 217.60 My a 902 1786 223.25 ‘908 ' ete 907 1809 226.125 " a 921 1828 228.50 uw 235 Vv 225y a of Trend) ot Je 231 (i) Least Square Method t-6 xaf65 £ ¥ 0.5 ie [ Trend value 4 210 —~*Y a xy - + 2 191 -9 121 = 2310 200.88 3 216 = at -1719 203.66 4 200 = 49 1512 206.44 25 ~ 1000 209.22 5 218 = 6 197 : bd - 654 212.00 7 230 4 : -197 214.78 fe 211 ‘ a 230 217.56 $ 9 633 220.34 246 a Bn e 25 1230 223.12 i as ma 49 1505 225.90 s 20 9 at 2115 228.68 12 S "1 421 2475 231.46 baie 2X=0 Ex =572 | ExX¥=796 | Let ¥= a+ hX be the trend equation, where a and h he ara calontated from the normal equations. EY=Na+bEX 0) EXY =aEX+bEX* Putting respective values from the above tables, we get 2594= 12a+0 or Again 796=0 + b(572) o «b= 28 21.39 572 <. Trend equation is ¥ = 216.17 + 1.39% AG For finding trend values we put X=~ 11, ~9, ..... in order and the corresponding trend ‘values have been shown in the last column of the above table Euample 34, From the given data, compute ‘trend’ and short-term fluctuations by moving average method (Periodicity = 4 years) 7980 | 1901 | 1982 | 1983 | 1984 | 1985 | 1986 | 1987 | 1988 | 1989 | 1990 525 | 470 | 515 | 550 | 575 | 560 680 | 620 | 645 ) Yea Production: 462 | 510 (in Qtls.)