🔗 GitHub
How to beat Sachin’s record of 100 100s!?
A data cleaning and visualisation was performed on a dataset containing cricketers with most runs in all formats. Through correlation, useful features were extracted and different types of regression models were trained to accurately predict the number of centuries a batter can score at the end of his career.
Columns
Player:- Player Name
Span :- Total time period
Mat :- Total Matches played by player
Inns :- Total innings
NO :- Total Not out
Runs :- Total runs
HS :- Highest Run in a innings
Ave :- Average of batting
BF :- Total Ball faced
start year :- Starting Year \
end year :- Ending Year
no of years played :- Experience
Technology used: Pandas, Matplotlib, Seaborn and Scikit-Learn
Heatmap
Observations through correlations
- Mat,Inns,Runs, Ave, BF,50,4s have high correlation with 100’s
- Therefore, in order to score more centuries all you have to do is play more innings, maintain good average and deal in 4s (6s have a very low correlation!).
- Filtering the features further - Inn, Ave and 4s
Paiplot on filtered dataset
The filtered dataset was trained with Multiple Linear Regression, Support Vector Regression, Decision Tree Regression and Random forest regression.
The r2 scores are as follows:
| Type of model | r2 score |
|---|---|
| Multiple Linear Regression | 0.83 |
| Support Vector Regression | 0.46 |
| Decision Tree Regression | 1 |
| Random forest regression | 0.93 |
Clearly the decision tree regression model wins!
Conclusion
In order to beat Sachin’s record of 100 100s, one should:
- Play as many innings as possible
- Maintain a good average
- Deal in a lot of fours
and just hope that they can beat ‘God’s’ record!