Root Mean Squared Error

A typical way of measuring the difference between observations and results from a predictor.

The formal definition is:

0

For $N$ observations $Y=\{y_1, \dots,y_N\}$ we can express it as:

1

Example

import numpy as np

X = 2 * np.random.rand(1000,1)
X_b = np.c_[np.ones((1000,1)), X]

Y = 1 + 2.5 * X + np.random.randn(1000,1)