Table of Contents

Enum StepwiseMethod

Namespace
AiDotNet.Enums
Assembly
AiDotNet.dll

Specifies the direction of feature selection in stepwise regression and other statistical models.

public enum StepwiseMethod

Fields

Backward = 1

Starts with all features and removes them one at a time based on their lack of statistical significance.

For Beginners: This is like starting with all players on a team and removing the least valuable players one by one.

How it works:

  1. Start with a full model (all features/variables included)
  2. Try removing each feature one by one
  3. Remove the feature that hurts the model the least
  4. Repeat steps 2-3 until removing any remaining feature would significantly harm the model

Advantages:

  • Less likely to miss important interactions between features
  • Can be better when features are correlated with each other
  • May find more complex relationships in the data

This method is useful when you suspect that combinations of features might be important, or when you're not sure which features to exclude.

Forward = 0

Starts with no features and adds them one at a time based on their statistical significance.

For Beginners: This is like building a team by starting with no players and adding the best available player one at a time.

How it works:

  1. Start with an empty model (no features/variables)
  2. Try adding each available feature one by one
  3. Add the feature that improves the model the most
  4. Repeat steps 2-3 until no more features significantly improve the model

Advantages:

  • Tends to create simpler models with fewer features
  • Often faster when you have many potential features
  • Easier to understand which features are most important

This method is useful when you have many potential features and want to build a streamlined model with only the most important ones.

Remarks

For Beginners: Stepwise methods help AI decide which information is important to consider.

Imagine you're trying to predict house prices. There are many factors that could affect the price: square footage, number of bedrooms, location, age of the house, etc. But using too many factors can make your model complicated and less accurate.

Stepwise methods help you decide which factors (called "features" in AI) to include in your model. They work by either starting with nothing and adding important features one by one, or starting with everything and removing less important features one by one.

Think of it like packing for a trip: you can either start with an empty suitcase and add only what you need (Forward), or start with everything you own and remove what you don't need (Backward).

Edit this page