Linear Regression

A linear model is an attempt to understand the target variable as a linear equation of the features. In other words, a linear model predicts the target(dependent) variable as a weighted sum of the independent variables. Linear regression models are very widely used by statisticians, computer scientists and other people who tackle quantitative problems.

The assumption that all relationships to a target variable are linear and that changing a feature input does not affect the other inputs makes one skeptical to use linear models. That being said, they are one of the oldest and traditional methods of making predictions with high Explainability.

The learned relationships in a linear model can be written for a single row/instance as follows

The equation format makes it very easy to understand how our prediction varies with change in every single feature input. In the coming sections, we will dive deeper into the assumptions of this model, the implementation & interpretation. Links to the Python notebooks are attached towards the end of every section for users.

Last updated