Tiêu đề Regression-2.pptx ✅

MACHINE LEARNING Dr.S.SAUDIA Assistant Professor CITE, M.S.University # REGRESSION TYPES AND APPLICATIONS
# REGRESSION TYPES Regression concepts get applied in supervised kind of Machine Learning Algorithms. Based on correlation between variables, the number and nature of independent variables considered in the regression model, feature engineering techniques adopted etc., the Regression algorithms are mainly classified as: 1. Linear Regression 2. Multilinear Regression 3. Polynomial Regression. Optimized Regression algorithms in literature are: 1. Ridge Regression 2. Lasso Regression 3. Elastic Net Regression. Course Teacher: Dr.S.Saudia, AP, CITE, MSU
# 1.LINEAR REGRESSION In the linear regression, the model obtained after training relates a dependent variable and a single independent variable. Initially correlation between the dependent variable and all the independent variables in the training dataset are found individually. Whichever independent variable produces largest correlation [negative or positive], that independent variable is fixed as the predictor. The parameters [m and c] of the best fit Regression line [y=mx+c] is arrived by minimizing the error between actual value of the dependent variable and the predicted value using Least Squares Technique. Residual analysis is already explained in lecture 14. Course Teacher: Dr.S.Saudia, AP, CITE, MSU
# LINEAR REGRESSION -PYTHON FUNCTIONS In Python the linear regression function is found in the sklearn library. It is available as the ‘LinearRegression’ class in its package ‘linear_model’. The linear regression model can be evaluated using Mean Square Error or R2 score. The functions for these metrics found in the sklearn library. They are available as ‘mean_squared_error’, ‘r2_score’ functions in the ‘metrics’ package of ‘sklearn’ library. Course Teacher: Dr.S.Saudia, AP, CITE, MSU