I am using a slightly modified version of the confidence_ellipse() function given in this example on the Matplotlib documentation. The only difference is I am returning the ellipse variable itself rather than "return ax.add_patch(ellipse)". I generated a random multivariate gaussian dataset for testing using this code: data = np.random.multivariate_normal(mean=(1,1), cov=[[0.4, 9],[9, 10]], size=1000) And here is the code I used to generate the ellipse and plot the data: fig, ax = plt.subplots(figsize = (5,5)) ellipse = confidence_ellipse(data[:,0], data[:,1], n_std = 2, edgecolor = 'green', facecolor = 'none') ax.add_patch(ellipse) ax.scatter(data[:,0], data[:,1], s=5, color = 'green', alpha=0.5) ax.set_xlabel('x') ax.set_ylabel('y') This returns something that generally looks like this: I am trying to figure out which of these points fall within the ellipse. I am using the contains_points() method to do this for the entire dataset: ellipse.contains_points(data) However, this shows that none of the points are inside the ellipse. So I tried transforming the dataset into the coordinate system of the ellipse: t = ellipse.get_transform() ellipse.contains_points(t.transform(data)) This doesn't seem to fix the issue either. It's showing that only 4-5% of the points fall within the ellipse, which is obviously incorrect. What am I doing wrong here? Any help is much appreciated! Continue reading...
