Filling zeros in time series data

fill_zeros

import numpy as np 
import pandas as pd

from datetime import datetime

# make an example
np.random.seed(0)

item = np.random.choice(['A', 'B'], 10)
year = np.random.choice([2016, 2017], 10)
month = np.random.choice(range(1, 13), 10, replace=False)
order = np.random.randint(low=1, high=10, size=10)

df = pd.DataFrame({'item': item,
                  'year': year,
                  'month': month,
                  'order': order})
                  
# create index based on item and year_month
df['year_month'] = df.apply(lambda row: datetime(row['year'], row['month'], 1), axis=1)
df = df.set_index(['item', 'year_month'])
df = df.drop(['year', 'month'], axis=1)

# create new index 
item_index = df.index.levels[0] 
date_index = pd.date_range('2016/1/1', periods=24, freq='MS')
iterable = [item_index, date_index]

new_index = pd.MultiIndex.from_product(iterable, names=['item', 'date'])

# reindex, if NaN, fill 0
df = df.reindex(new_index, fill_value=0)

# unstack
new_df = df.unstack('item', fill_value=0)

Very nice! 👍