statsmodels|sklearn|matplotlib|pandas|numpy

ols|LinearRegression|LeaveOneOut|Kfold|train_test_split

Task : build a simple linear regression model to predict delivery time using sorting time

Summary : I build a simple linear regression model using :

• ols method
• LinearRegression

I used ols to understand some basic terms like R-squared, adj. R-squared, AIC, significance of p-value. Further I build model using LinearRegression. The techniques I used in building model are - LeaveOneOut, Kfold, train_test_split to compare all these with the same dataset.

Conclusion :

• Model build using ols has R-squraed = 68% and LinearRegression have accuracy approximately in the line of 60's.
• All the three techniues - LeaveOneOut, Kfold, train_test_split give approximately same accuracy except KFold (33%).
• Accuracy remain in the line of 60's because we use a very small dataset (21 records).