Merge and Sum Pandas Rows

Question

I have a Pandas dataframe that looks like this:

date      item       amount
201901    Apple      1.03
201901    Potato     1.04
201901    Orange     1.00

I'm trying to find the sales of fruits and vegetables by month:

date      item        amount
201901    Fruit       2.03
201901    Vegetables  1.04  

What's the best way to do this? I'm familiar with df.groupby(['date','item'])['amount'].sum() , but this does not conditionally combine the fruits and veggies.

One way is to create another column type based on the value in item and then group on that; is there a better way?

Answer 1

As Manakin said, you need to manually classify your items.

Build a mapping dictionary with item : category pairs and pass it to series.map or series.replace .

map will change all the items that are in the dictionary, and fill with NaN otherwise. replace will find and replace all matching items and replace them, but will leave items not in the dictionary keys as they are (eg if the dataframe contains 'brussel sprouts' but that key is not in the dictionary the it will leave it as the item name). It is up to you to decide which behavior you need.

Here's an example with series.map :

categories = {'Apple': 'Fruit', 'Potato': 'Vegetable', 'Orange': 'Fruit'}

df['category'] = df['item'].map(categories)
summary = df.groupby(['date', 'category'])['amount'].sum().reset_index()
print(summary)

Output

     date   category  amount
0  201901      Fruit    2.03
1  201901  Vegetable    1.04

Answer 2

You should probably have 2 lists or a dictionary of what you consider a fruit or a vegetable but when you do...

 mapping = {'Apple': 'Fruit', 'Potato': 'Vegetable', 'Orange': 'Fruit'}

This could give you what you want without a need to add a column , computing the grouping in the fly:

 def grouper(row):
     return row['Item']

 group_earnings = (df.groupby(grouper))['amount'].sum().reset_index()