### Isotonic Regression¶

• Fits a non-decreasing function to 1D data. If minimizes $\sum_i w_i (y_i - \hat{y}_i)^2$ where the weights $w_i$ are all positive, and both x and y are real values.
• increasing changes the constraint to $\hat{y}_i \ge \hat{y}_j$ when $X_i \le X_j$. increasing="auto" chooses a constraint based on Spearman's rank correlation coefficient (Wikipedia).