Build an Israeli Voter

Hi -

So I have been learning about and testing with machine learning. Specifically, Random Forest Classifications. I have been doing this on the Israeli Election Study. It contains the responses and views of hundreds of Israelis, from different backgrounds - which makes it a great opportunity to test out various models.

Specifically - I have made a tool that projects how a Israeli voter of a particular demographic will vote. It is around 80% accurate (AUC). I guess the other 20% is something to do with free will (or a small sample size)...

To do this, I have broken up the sample into the Arab and Jewish Sector, and then by various grouped demographic sections.

The findings seem to fairly closely match up with my mapping of Israeli election results from 1992-2022. This is a good sign in terms of accuracy too.

In terms of findings the main I can find initially is that nationalism tends to be more bottom up from Jewish Israelis (religious, more deprived), but top down from Arab Israelis (secular, less deprived). The former I am much more familiar with in Europe, but the latter tends to hold more true across the Middle East.

I am super excited to share this with you - I hope you find it as interesting as I do. I'd like to thank Chris Terry for his help in making this model and teaching me about the basics of statistics.

Disclaimer: the results are not completely accurate, particularly for low sample size groups. Do not use for serious purposes.

As ever, the code is on Github.