Linear Regression

Supervised learning algorithm that formulates a linear line in order to predict the target variable

Baseline Equation

$$r={\theta }_0+{\theta }_1{x}_1+{\theta }_2{x}_2+....$$

Training

• Bias denoted with ${\theta }_0$ and Weights denoted with ${\theta }_i$ are trained and used to formulate the line equation
• The weights and bias are learned through the Gradient Descent Algorithm

Non Linear Relationships

• If we come across relationships that are not linear such as $x^2$ or $x^3$, we convert those relations into linear features such as creating a new dimension for $x^2$ while keeping $x$