Multivariate Normal Modeling in Python Using Numpy

In this video I will go over how to write your own class to to fit data to a multivariate normal distribution. This will be useful for future videos when I cover unsupervised learning topics such as clustering and anomaly detection.

[@chonky_ollie]

You are a legend. Thank you!

[@snesh93]

Can you provide more insight on the probability plot. I kinda felt it was a bit quick and couldnt apprehend it fully.

[@cspythonforscience]

Yes sorry I sort of went though it fast or otherwise the video was going to be really long. For contour plots I recommend checking this link out. It is where I learned how to make these plots. Hope this helps!
jakevdp.github.io/PythonDataScienceHandbook/04.04-density-and-contour-plots.html

[@brianmunson4861]

Can you explain the ensue portion around 6:36 of the video? Awesome tutorial by the way

[@cspythonforscience]

Are you talking about the einsum method?

[@brianmunson4861]

@@cspythonforscience Yes, I would like an explanation of the selection of the unit vectors used and why. The data set I am trying to apply this to is a 2d numpy array (256,256). It seems that either my data needs to be restructured, or the functions need to be altered. Thank you for your time

[@cspythonforscience]

@@brianmunson4861 Ok I will try to upload a new video soon talking about it, hopefully within the next week. Here is the article I read that really explained how to do einsum, at least for me. rockt.github.io/2018/04/30/einsum It makes doing dot products, transpositions, and summations with higher dimensional array much easier. Thank you for the kind words!