ampli/py/projprocess.py

# This file simply takes future kwh data for the thousand previously 
# sampled ICPs and calculates new aggregated measures for each cluster

import pandas as p

# Load new data
newdat = p.read_pickle("../data/2018-proj-sample.pkl")
print(newdat.info())

# Get cluseters dataframe
clusters = p.read_pickle("../data/9-clusters.pkl")
print(clusters)
print(clusters.info())
clusters = clusters.drop(['read_time', 'kwh_tot'], 
                         axis = 1).drop_duplicates().reset_index(drop = True)

# Join dataframes
newdat = clusters.set_index('icp_id').join(newdat.set_index('icp_id'), how = 'left').reset_index()

print(newdat)
print(newdat.info())

# Aggregate median and mean (only really want mean)
newagg = newdat.groupby(['read_time', 'cluster']).agg({
        'kwh_tot': ['median', 'mean']
})
newagg.columns = ['_'.join(x) for x in newagg.columns.ravel()]
newagg = newagg.reset_index()

print(newagg)
print(newagg.info())

# Save data
newagg.to_pickle("../data/9-proj-agg.pkl")