Hands-On ROS for Robotics Programming

Regression

Regression tries to find the curve that best fits a cloud of points, and it has been described in detail with the case of the prediction of house prices. In such a case, we have been talking about a linear dependency, but the algorithm can be generalized to any kind of curve that can be represented as a sum of dot products between coefficients (weights) and independent variables (features), that is, a polynomial. A common case is that of a term that is the square of a feature. In this case, the curve is a parabola and, mathematically, can be expressed as follows:

y = W1 * x + W2 * x² + W3 * 1

Let's review this with a real-life example. Given an independent variable, the years of experience of a candidate, we wish to predict what ...