A Multiple linear regression (MLR) model computes the dependent variable Y using a linear combination of independent variables X_k.
The regression model requires that input variables (X) explain the output variable (Y), and furthermore that each input variable adds different information. The following methods were tested:
- Forward selection
- Backward selection
- Stepwise selection (use the previous ones)
by adjust of R-squared, Variance Inflation Factor (VIF), p-value, etc.
Dependences:
python - Pandas
python - NumPy
python - Matplolib
python - Statsmodels
Page source: