Applied Linear Algebra Project: Mid-Sem Presentation

You might also like

Download as pptx, pdf, or txt
Download as pptx, pdf, or txt
You are on page 1of 11

Applied Linear Algebra Project

Mid-Sem Presentation
 Khush Soni
 Divyashree Jadeja
 Aadya Chinubhai

Topic :
Using Ordinary Least Squares to predict
prices of used Cars.
Background:

10/28/2021 | 2
Motivation
 We were always interested in Machine Learning and AI. We were always fascinated on How does

AI work?

 So, this was the perfect chance to do a Machine Learning project.

 Before going into complex AI research projects, it was important to master basic Machine

Learning Algorithms like Least Squares.

 Also, since we can’t use any ready libraries like scikit-learn or Tensorflow , this project would

provide us with a lot of in-depth knowledge in the field.

 Also, this was the perfect project to give a Teamwork experience and mimic real life situations

like Jobs or AI Teams.


Problem
statement:
linear regression is used to display the connection between a dependent
variable and one or more independent variable.
 Our challenge is to determine the value of m and c, that provides the
minimum error for the given the dataset. the Least Squares method will help
us in that.
 By using least square method we need to predict the prize of cars. The
regression line will contain values which prediction values for prize of cars
where as the points on graph are the actual values from data set. for those
prediction to be as close as real values we need square of the errors as small
as possible.
Reproduced Work:
 The approaches to fitting straight lines to collections of x−y data pairs can be broadly
grouped into two categories: the “standard” least-squares methods in which the
distances between the fitted line and the data in the y-direction are minimized, and
the “bivariate” least-squares methods in which the perpendicular distances between
the fitted line and the data are minimized.
 From the given set of a data we have many point and we need a
line which is as close as possible to all points, and a similar
number of points above and below the line. This line is called
regression line. 
 we will calculate the line using Least Squares Regression.
 Our target is to calculate the values m - slope and c(y intercept) in the
equation of line.

 Our challenge is to determine the value of m and c, that provides the minimum
error for the given the dataset. the Least Squares method will help us in that.
10/28/2021 | 6
 Values of m and c changes as position of regression line changes.
Finding the Error:
 We need to find the error first and than we will try to minimize error. So that we can fit line
such that it is at least distance from each point.
 Calculation error means calculating difference between actual value and prediction value.
 we will be using the Quadratic Loss Function to calculate the error. for the points below the
regression line y — p value of error is negative. We square this value so we do not get any negative
value.
 After finding loss function we need to minimize it. That is done by finding the
partial derivative of L(x) and equating it to 0 and then finding an expression
for m and c.

 This is called the Least Squares method.

Here,
x̅ = the mean of all the values in the input X 
ȳ = the mean of all the values in the desired output Y
Contributions:

Divyashree Jadeja :

Dataset Cleaning

Documenting work e.g.: Making ppt/ Tracking progress

Khush Soni
Making ML model in python, background.

Aadya Chinubhai

Making ML model in python and testing it.


Plan of Action:

Helping in building
Making ML Model Model and then
Dataset Cleaning
using Numpy testing it to get
accurate output.

 In data cleaning, We remove unnecessary components from the dataset so that the model is more precise. Example :
Car prices depend only on some factors so we remove the unnecessary factors so that they don’t affect the model.
 Also, We can only feed numbers to the model. So, converting words to numbers is also done in this step.

 Making ML models is done next :


Models essentially give the dependence of Output variable on the Input variables. The coding for it is done in this
step. Here is where Linear Algebra comes into play.

Testing the model :


 Here we feed new data to the model and check how accurate it is.

You might also like