Python pandas: calculate revenue from price and quantity

Question

I have a dataframe that looks like the following:

df
Out[327]: 
        date  store property_name  property_value
0 2013-06-20      1         price             101
1 2013-06-20      2         price             201
2 2013-06-21      1         price             301
3 2013-06-21      2         price             401
4 2013-06-20      1      quantity            1000
5 2013-06-20      2      quantity            2000
6 2013-06-21      1      quantity            3000
7 2013-06-21      2      quantity            4000

I would like to calculate revenue for each date, for each store then add that to the bottom of the dataframe. For example, for 2014-06-20, for store#2: revenue=201*2000 = 402000.

Below is my code but I know it's not efficient for larger dataframe:

import pandas  as pd

dates = df['date'].unique()
stores = df['store'].unique()
df_len = len(df)
for date in dates:
    for store in stores:       
        mask_price = (df['date']==date) & (df['store']==store) & (df['property_name']=='price')
        mask_quantity = (df['date']==date) & (df['store']==store) & (df['property_name']=='quantity')
        price = df.loc[mask_price,'property_value'].iloc[0]
        quantity = df.loc[mask_quantity,'property_value'].iloc[0]

        df.loc[df_len,'date'] = date
        df.loc[df_len,'store'] = store
        df.loc[df_len,'property_name'] = 'revenue'
        df.loc[df_len,'property_value'] = price*quantity

        df_len=df_len+1

Thank you in advanced for your help :)

Answer 1

This is one way.

price = df[df['property_name'] == 'price'].set_index(['date', 'store'])['property_value']
quantity = df[df['property_name'] == 'quantity'].set_index(['date', 'store'])['property_value']

rev = (price * quantity).reset_index().assign(property_name='revenue')

df = pd.concat([df, rev], ignore_index=True)

Explanation

• Derive price and quantity dataframes via slicing, index by date and store .
• Calculate rev via price * quantity on index; add property_name columns.
• Concatenate along axis=0 by default (index).

Result

          date property_name  property_value  store
0   2013-06-20         price             101      1
1   2013-06-20         price             201      2
2   2013-06-21         price             301      1
3   2013-06-21         price             401      2
4   2013-06-20      quantity            1000      1
5   2013-06-20      quantity            2000      2
6   2013-06-21      quantity            3000      1
7   2013-06-21      quantity            4000      2
8   2013-06-20       revenue          101000      1
9   2013-06-20       revenue          402000      2
10  2013-06-21       revenue          903000      1
11  2013-06-21       revenue         1604000      2

Answer 2

Another way of doing it:

prices = df[df['property_name'] == 'price']
quantities = df[df['property_name'] == 'quantity']

res = prices.merge(quantities,on=['date','store'],how='left')
res['property_value'] = res['property_value_x']*res['property_value_y']
res['property_name'] = 'revenue'
res = res[['date','store','property_name','property_value']]

res = prices.append([quantities,res])

Same logic as first answer here:

1. Separate prices and quantities
2. Merge both tables using date and store as a key
3. Compute wanted column in a third table
4. Concatenate everything

Hope that helps.