A contest about measuring molecular measurements. My hope was to reduce the problem to a geometric problem and then use distance measures and weighted measures to arrive at a result.
Also note that I chose to ignore the update in rules to allow the use of the extended dataset. The hosts of the competition were unhappy about the development and I wanted to respect their original intentions. I wanted to see how the challenge would be originally and what I could achieve with the original dataset.
My features included checking the directions of the position vectors, and distance measurement.
Some of the notebooks I was influenced by: (1). https://www.kaggle.com/marcelotamashiro/lgb-public-kernels-plus-more-features (2). https://www.kaggle.com/criskiev/distance-is-all-you-need-lb-1-481 (3). https://www.kaggle.com/mykolazotko/3d-visualization-of-molecules-with-plotly (This kernel is particularly important to the methods I think will be conducive to solving this)