Wow, good thing you decided to clean out all that. You found some real problematic artifacts in there. Some doofus put the timestamps for the panda birthdays in excel and you began optimizing on one hundred year old pandas before you found the mistake! There’s nothing wrong with just using a regression model, especially with well cleaned data. It’s fitting, to something, you’ve got a score now and can run some optimization metrics to figure out which pandas should have Netflix and chill dates with which other pandas. It is all very explainable and you can even track your uncertainty. Maybe there were some more complicated interactions going on, or maybe this was all you needed. Your feature space was selected with some careful pruning processes you didn’t want to think about too much. You even did some K-fold cross validation and found that the fit did not change much. What do you do?
You cleaned that data before doin a little of that Regression
