My baseline solution for AICROWD's Learning to smell competition.
I use precomputed fingerprints from PubChem database, which are collected at pubchem_fingerprints.csv
file. The script download_data_from_pubchem.py
can be used for downloading them (might be slow, because it downloads data from remote server).
- Download files from competition and save them to
data
directory. - File
pubchem_fingerprints.csv
is already precomputed - it is output frompython download_data_from_pubchem.py
run. - You can run
baseline_solution.ipynb
with jupyter notebook.
You can also read my Medium post about this solution both with some ideas what to do next.