Feature Selection - Machine Learning
Understanding Feature Selection
- It is one of the key stages in building Machine Learning Models.
- All input variable may not useful to build model as each variable may not have enough predicting power.
- Simple model is preferred, If a Model with less Columns is having accuracy as same as another with model more number of then we prefer Model with less less number of columns columns
- It saves Time & Resource utilization.
Feature Selection Techniques
The following are a few methods we use them very often.
- Variance Threshold
- Multicollinearity
- Chi-Square Test
- MSE - Univariate
- Forward Selection
- Backward Elimination
- L1 Normalization - LASSO
- L2 Normalization - RIDGE]
- Gini Index
- * Decision Trees
- * XBGBoost
