Implementing Gaussian Naive Bayes from Scratch

preview_player
Показать описание
In this video, we implement Gaussian Naive Bayes model from scratch.

#ai #softwareengineering #programming #stylepoint #naivebayes

Chapters
0:00 - Intro
0:40 - Thinking about the API
1:35 - Implementing the `fit` method
5:14 - Implementing the `predict` method
9:37 - Testing
9:57 - Outro
Рекомендации по теме
Комментарии
Автор

two comments:
1: the api is unusual - there is no need to save the training data on the instance, those should be args to `.fit` rather that `__init__`
2: it's numerically more stable to use log-likelihoods (and sums instead of products)

grosucelmic
Автор

I have added the timestamps for the video.
Hope this will be helpful (:

davidoniani
Автор

This was really well done. Straight to the point but still elaborating on details people may have forgotten. My one suggestion: before testing the full dataset, it would've been nice to step through the finished code with a toy dataset and see the result of each step. Jumping right into performance metrics made the end seem greater than the sum of its parts

whaley
Автор

Really enjoyed the video, nice work!
Perhaps in the description it would be useful to include a link to a good quality source for those hoping to understand the underlying theory of the algorithm in more depth.

joshwapiano