R

Introduction Suppose we observe pairs $(X_1, Y_1), \ldots, (X_n, Y_n)$ where we want to estimate the regression function $$ m(x) = \mathbb{E}[Y \mid X = x]. $$The classical Nadaraya–Watson estimator is defined as $$ \hat{m}_n(x) = \frac{\sum_{i=1}^n K_h(x - X_i)\, Y_i}{\sum_{i=1}^n K_h(x - X_i)}, $$where $K_h(u) = h^{-1} K(u/h)$ is a kernel with bandwidth $h > 0$. The Recursive Version In my thesis, I studied a recursive variant suited for streaming data. Define the sequence of estimates: ...