just few lines get 0.73 lb 'n' 0.74 p lb,
import pandas as pd
df = pd.read_csv( "./SampleSubmission.csv" )
df['crop_type_cocoa'] = 0.75
df['crop_type_oil'] = 0.125
df['crop_type_rubber'] = 0.125
df.to_csv( "./submission.csv", index =False )
I trained submission model with balanced weights on loss but private data with highly unbalance with log_loss to evaluation ( which don't care about minor class )