How I won my 12th bronze medal on Kaggle’s April 2022 tabular competition

This morning I was quite pleased to learn that I had earned a bronze medal on Kaggle’s April 2022 tabular competition that I had only just entered the day previously. This is a time series prediction problem where a particularly binary state is predicted based on various sensor readings.

I found the competition question to be a bit difficult because the train labels are found in a separate dataframe and they must be mapped to the train dataframe based on the sequence, which is a common column:-

In addition, the submission was based on the sequences of the test dataframe. I was given no guidance on how to calculate this, so I wrote my own code to solve the problem:-

Because of limitations of my computer, I had to use sklearn’s LogisticRegression method to make the…