LEARNING OBJECTIVES

 Describe the importance of forecasting to the value chain.

 Explain basic concepts of forecasting.
 Explain how to apply simple moving average and exponential smoothing models.
 Describe how to apply regression as a forecasting approach.
 Explain the role of judgement in forecasting.

What Is Forecasting?

Forecasting is a technique that uses historical data as inputs to make informed estimates that are predictive in
determining the direction of future trends.

Businesses utilize forecasting to determine how to allocate their budgets or plan for anticipated expenses for an
upcoming period. This is typically based on the projected demand for the goods and services offered.

How Forecasting Works

Investors utilize forecasting to determine if events affecting a company, such as sales expectations, will increase or
decrease the price of shares in that company. Forecasting also provides an important benchmark for firms, which need a
long-term perspective of operations.

Forecasting Techniques

In general, forecasting can be approached using qualitative techniques or quantitative ones. Quantitative methods of
forecasting exclude expert opinions and utilize statistical data based on quantitative information. Quantitative
forecasting models include time series methods, discounting, analysis of leading or lagging indicators, and econometric
modeling that may try to ascertain causal links.

Time series patterns

Time series data is a sequence of data points that measure some variable over ordered period of time. It is the fastest-
growing category of databases as it is widely used in a variety of industries to understand and forecast data patterns. So
while preparing this time series data for modeling it’s important to check for time series components or patterns. One
of these components is Trend.

In describing these time series, we have used words such as “trend” and “seasonal” which need to be defined more
carefully.

Trend

A trend exists when there is a long-term increase or decrease in the data. It does not have to be linear. Sometimes we
will refer to a trend as “changing direction”, when it might go from an increasing trend to a decreasing trend. There is a
trend in the antidiabetic drug sales data shown in Figure 2.2. Trend usually happens for some time and then disappears;
it does not repeat. For example, some new song comes, it goes trending for a while, and then disappears. There is
fairly any chance that it would be trending again.

Seasonal

A seasonal pattern occurs when a time series is affected by seasonal factors such as the time of the year or the day of
the week. Seasonality is always of a fixed and known frequency. The monthly sales of antidiabetic drugs above shows
seasonality which is induced partly by the change in the cost of the drugs at the end of the calendar year.

Many time series display seasonality. By seasonality, we mean periodic fluctuations. For example, retail sales tend to
peak for the Christmas season and then decline after the holidays. So time series of retail sales will typically show
increasing sales from September through December and declining sales in January and February.
Cyclic

A cycle occurs when the data exhibit rises and falls that are not of a fixed frequency. These fluctuations are usually due
to economic conditions, and are often related to the “business cycle”. The duration of these fluctuations is usually at
least 2 years. One example of a cyclical pattern, the business cycle, is from macroeconomics. Over time, economic
expansions are followed by economic recessions followed again by economic expansions. There is not perfect
regularity in the business cycle, as expansions and recessions differ in length. Nevertheless, this process has repeated
itself over and over through time.

Many people confuse cyclic behaviour with seasonal behaviour, but they are really quite different. If the fluctuations are
not of a fixed frequency then they are cyclic; if the frequency is unchanging and associated with some aspect of the
calendar, then the pattern is seasonal. In general, the average length of cycles is longer than the length of a seasonal
pattern, and the magnitudes of cycles tend to be more variable than the magnitudes of seasonal patterns.

 The monthly housing sales (top left) show strong seasonality within each year, as well as some strong cyclic
behavior with a period of about 6–10 years. There is no apparent trend in the data over this period.
 The US treasury bill contracts (top right) show results from the Chicago market for 100 consecutive trading days
in 1981. Here there is no seasonality, but an obvious downward trend. Possibly, if we had a much longer series,
we would see that this downward trend is actually part of a long cycle, but when viewed over only 100 days it
appears to be a trend.
 The Australian quarterly electricity production (bottom left) shows a strong increasing trend, with strong
seasonality. There is no evidence of any cyclic behavior here.
  The daily change in the Google closing stock price (bottom right) has no trend, seasonality, or cyclic behaviour.
There are random fluctuations which do not appear to be very predictable, and no strong patterns that would
help with developing a forecasting model.

What is forecast accuracy and forecast error?

One way to check the quality of your demand forecast is to calculate its forecast accuracy, also called forecast error. The
forecast accuracy calculation shows the deviation of the actual demand from the forecasted demand. If you can
calculate the level of error in your previous demand forecasts, you can factor this into future ones and make the
relevant adjustments to your planning.

In this post we show you how to measure the accuracy of your forecasts, by calculating forecast error, and then discuss
why it’s important to do so.

Forecast accuracy/forecast error calculations

There are a number of formulas that inventory planners can use to calculate forecast accuracy / forecast error. These
range from the fairly simple to the quite complex. Two of the most common forecast accuracy / error calculations are
MAD – the Mean Absolute Deviation and MAPE – the Mean Absolute Percent Error.

Let’s take a closer look at both

1. MAD forecasting calculation

A common way to work out forecast error is to calculate the Mean Absolute Deviation (MAD). This shows the deviation
of forecasted demand from actual demand, in units.

The MAD calculation takes the absolute value of the forecast errors (difference between actual demand and the
forecast) and averages them over the forecasted time periods. ‘Absolute value’ means that even when the difference
between the actual demand and forecasted demand is a negative number, it becomes a positive. So 25 divided by 4 is
6.25.
The MAD calculation works best when using it on one product, as the demand error is not proportional. If you use it on
items with different volumes, the result will be skewed by those with heavier volumes.

2. MAPE forecasting calculation

Another fairly simple way to calculate forecast error is to find the Mean Absolute Percent Error (MAPE) of your forecast.
Statistically MAPE is defined as the average of percentage errors. The MAPE formula consists of two parts: M and APE.
The formula for APE is the difference between you actual and forecasted demand as a percentage:

Since MAPE is a measure of error, high numbers are bad and low numbers are good.

There are other forecast accuracy calculations that you can use, but make sure you find the most appropriate method
for your needs, as it’s important to understand how accurate your forecasting is for a few reasons that we will now
discuss.

Using forecast error data for better demand predictions

Once you have your forecast error calculations, you need to ensure you act on the data. Smart inventory planners will
use their forecast error stats to refine their forecasting processes and improve overall forecasting accuracy. More
accurate forecasts will then help improve their inventory purchasing and planning.

Here are a number of ways this can be done:

1. Mitigate the risk of future forecasting accuracy: The forecast error calculation provides a quantitative estimate of the
quality of your past forecasts. If you can calculate the level of error in your previous demand forecasts, you can factor
this risk into future forecasts. If you can determine how uncertain a forecast is for a given future business period, you
can make the necessary adjustments to your inventory management rules, such as increasing safety stock levels and
adjusting re-order points to cover the uncertain periods of demand.

2. Prioritise questionable forecasts: Identifying and prioritising items with a high forecast error allows to you give them
dedicated attention. You can closely monitor their future demand and adjust stock levels accordingly.

3. Refine and improve forecast accuracy: If you consistently see high forecast error rates this is an indication that the
demand forecasting technique you’re using needs to be reviewed and improved.