Give a dataset $\mathcal{D} = \{(\mathbf{x}, y) | \mathbf{x} \in R^d, y \in R \}$, we can build a *linear* model to *predict* $y$ (label) given only $\mathbf{x}$ (features). Formally, a linear model can be written as $\hat{y} = f(x) = \mathbf{x}^T\beta$, where $\beta$ is parameter vector that we can tune and $\hat{y}$ is our … Continue reading Theory of Linear Regression