English: Fitting to noisy data is common in science and in machine learning. When a function - here a Gaussian function - is fit to noisy data, we want to know how good this fit is. One way is to plot the residuals, i.e., data y values minus predictions of fitted functions. Top panel shows noisy data in red plus a fit of a Gaussian function in green, while bottom panel shows a plot of the residuals in red plus a fit of a straight line (blue dashed) to the residuals. Note that the residuals are small in comparison to the values of the data (note that the two y scales are different), and the slope of the fit to the residuals is small - there is little trend in the residuals. Small residuals plus little or no trend in the residuals suggests that the Gaussian function is a good fit to the data.