Practical Data Science using Python - Linear Regression Model Evaluation and Optimization

The video tutorial covers feature selection techniques for optimizing linear regression models. It begins with dropping features based on probability values, then checks multicollinearity using Variance Inflation Factor (VIF). Recursive Feature Elimination (RFE) is introduced for automatic feature selection. The tutorial concludes with model testing, validation, and performance evaluation, achieving an R-squared score of 86% for the test data.

What is the threshold p-value used to decide which features to drop in the initial model optimization?

• 0.01

• 0.10

• 0.15

• 0.05

What does the Variance Inflation Factor (VIF) help to identify in a dataset?

• Data scaling issues

• Missing values

• Multicollinearity

• Outliers

At what VIF value is a feature typically considered for removal due to multicollinearity?

• 10

• 5

• 3

• 1

What is the main purpose of recursive feature elimination in model optimization?

• To improve data scaling

• To increase the number of features

• To automatically select significant features

• To reduce the dataset size

What is the adjusted R-squared value achieved after recursive feature elimination?

• 0.85

• 0.90

• 0.92

• 0.99

What is the purpose of adding a constant in the model testing phase?

• To remove outliers

• To include a bias term

• To improve accuracy

• To scale the data

What is the R-squared score achieved on the test data set?

• 92%

• 86%

• 82%

• 76%

Why is it important to match the test data features with the final model's features?

• To improve visualization

• To increase data size

• To reduce computation time

• To ensure model compatibility

What is the significance of testing residual normality in model evaluation?

• To validate OLS assumptions

• To reduce data size

• To check for multicollinearity

• To improve R-squared value

What is the mean value around which the residuals should be distributed for normality?

• -1

• 0

• 1

• 0.5