The Naive Bayes classifier is an interesting construction that uses conditional probabilities in order to attempt to determine the probability that an object is in a particular class. The example that this is used with is spam filtering, and this is apparently one of the methods that was once used for that task.

After completing my version of the voting information interactive bar chart, I realized that it was built off d3 version 3, and the current version is version 5. So I decided that I would go through and update the code so that it runs on the modern version. Things are still taking longer than I think they should, but I’m still in the learning phase, so I’m okay with that. Here is the link to the new version

It took several hours (and at least an hour more than it ought to have taken), but I took the interactive bar chart and modified it take advantage of all of the data that was available. So I now have my own version of the voting information interactive bar chart. I’ve added the feature that you can select the year (since that data was available in the original data set). In order to get myself to fully understand it, I basically rebuilt the entire thing from scratch, but basing it off the original model.

The KNN classifier is the one that I explained to my CS course a few semesters ago because it’s the most intuitive classification algorithm. It’s so intuitively simple that this always seems to be the first machine learning algorithm that’s presented.

I really got my first introduction to machine learning watching Andrew Ng’s CS229 lecture videos from 2008 or so. I probably watched it 2-3 years ago by this point. So this isn’t my first introduction to machine learning. It’s hard for me to think about how it would be reading this through for the first time, but I think it feels like it’s a nice, gentle introduction that avoids some of the more formal mathematical points.

I spent some time thinking last night, and I’ve decided that it’s probably good for me to start spending some time to shore up my web skills and dive into JavaScript and d3. It’s not that I’ve never done any web programming, but what I have is very basic. And I’ve never touched d3 (nor had I heard of it before seeing it in the Data Science from Scratch book). So I started by doing some reading and brute force copying just to get things going.

This data exploration is a good overview chapter, but since it’s doing the “from scratch” approach, I think it leaves a lot of tools on the table for the future. While I’ve appreciated seeing some of the other pieces from scratch, in this case it seems that it may hurt more than it helps.