Linear Regression
- 1. LINEAR REGRESSION I N P Y T H O N
- 2. Type of Supervised Learning Algorithm. Regression refers to searching for relationship between variables. Simplest type of regression, used Use Cases: predict sales, predict impact of change of predictor, forecast, risk assessment, check for trends when variables are "linearly" related WHAT IS IT?
- 3. Works on linear relationship of variables Example, lets have a small table: Here, y is a dependent variable, which depends on x. MATH BEHIND
- 4. Find the best fit line, which goes or is close to almost every point in the table (data) Let's suppose a line l passes through it. l can be defined as: It is the slope intercept form of line. m is slope and c is the intercept for the line which passes from (x,y) y = m . x + c GOAL
- 5. Let's suppose, the best fit line passes through x mean and y mean. (3, 3.6) is the mean point for the tabular data MATH
- 6. Slope, m of the line: If the line passes through (3, 3.6) and have slope m as 0.4, we can easily determine 'c' by subsituting the values and solving for 'c'. 3.6 = 0.4 (3) + c Thus, c comes to be 2.4 MATH 2
- 7. The equation of lines comes to be: This is our regression line. Y (pred) are the predicted y values corresponding to the x values y = (0.4) x + 2.4 MATH
- 8. We calculate the accuracy by R-Square method. 1 - R2 is the error HOW PERFECT IS IT? = = 0.3
- 9. Implementing the linear regression using scikit learn. SCIKIT LEARN