"No one is harder on a talented person than the person themselves" - Linda Wilkinson ; "Trust your guts and don't follow the herd" ; "Validate direction not destination" ;

March 10, 2016

Regression Basics

This post is on basics of Regression and Steps Involved. Linear Regression defines relationships between variables involved. We use it to identify relationships between variables.

Steps Involved
  • Plot line between Independent Variable in X Axis, Dependent Variable Y Axis
  • Identify if their positive or negative relationship (When X increases with respective to Y it is positive)
  • Plot a line that minimizes errors between estimates / actuals
Y = B0 + B1X (B0, B1 Derived Mathematically)
where B0 is Y Intercept, B1 is Slope

R Squared 

R Squared Verification 
  • How well regression line predicts actual values
  • Take Actual values (compute mean of them). Distance between actual value of mean will sum up to zero
  • Perfect fit R square equals 1

Standard Error of Estimates
  • Compare estimated values vs Actual Values
  • Distance between estimated and actual values

Correlation Coefficient
  • Fit the line
  • Remember slope +ve or -ve
  • Scatter along Y and X Axis
  • High Correlation means good fit

In next post we will look @ R Examples

Happy Learning!!!

No comments: