ridge regression calories prediction.pdf

Calories Burnt
Prediction using Ridge
regression and XGBOOST

1. Calories Burnt
Prediction using
Machine Learning
with Python

2. Sports and new
Technologies
TABLE OF CONTENTS
01
02 What Happens when you
exercise ?
Problematic
Introduction
Working method
03
04
Conclusion
RESULT
CRISP-DM
2

3. INTRODUCTION
Today’s Topic
01
3

4. “How can they measure burnt calories
while exercising ?”
PROBLEMATIC
02
4

5. Correlations between
calories burnt, exercise
and body type
SOLUTION
5

6. CRISP-DM
CRoss Industry Standard Process for Data Mining
03
6

7. CRISP-Dm
7

8. Data Understanding
8

9. Statistical measures about the data
9

10. Data Visualization
Mean
Median
Mode
10

11. Heatmap
The sample
correlation coefficient
(r)

●r>1 : highly correlated
●r = 0 : no correlation
●r<-1 : negative correlation

11

12. CRISP-Dm
12

13. Data preparation
13

14. CRISP-Dm
14

15. Modeling
15

16. Regression is a modeling task
that involves predicting a
numeric value given an input.
Linear
regression
Relationship between
dependent variable (Y) and one
or more independent variables
(X) using a best fit straight line
.
16

17. Ridge
regression
Technique used when the data
suffers from multicollinearity .
?
Ridge regression solves the
multicollinearity problem
through shrinkage the
coefficients for those input
variables that do not
contribute much to the
prediction task .
Including all parameters leads
to a high level of overfitting.
Find the coefficients that
minimize the sum of error
squares by applying a penalty
to these coefficients.
17

18. Ridge
regression
Parameter λ penalty term :

If λ is zero, this gives us the classical
regression equation.

The higher the Alpha values, the
greater the penalty ,therefore, the
size of the coefficients is reduced.
Better than the Least
Squares method when
there are too many
parameters.


18

19. Penalty term λ
19

20. Modeling
XGBoost
XGBoost is an efficient algorithm that can
be used for regression predictive
modeling.
20

21. Modeling
Testing set
Training set
70 %
30 %
21

22. CRISP-Dm
22

23. Evaluation
R Square
Mean Absolute Error
Mean Square Error
n = the number of
errors,
Σ = summation symbol
|xi – x| = the absolute
errors.
mae = metrics.mean_absolute_error(Y_test,
test_data_prediction)
Mean Absolute Error = 2.7159012502233186
23

24. XGBOOST
model = XGBRegressor()

model = Ridge(alpha=1.0)
Ridge
regression
24

25. conclusion
results
25