CODEWITHSUNDEEP
pythonpandasdataframe
joeDiHare
joeDiHare

Reputation: 679

Using one DataFrame as index for another DataFrame in pandas

Still trying to figure out how to perform operations with multiple DataFrames form pandas, in Python.

I have the following three dataframes (d1, d2, and d3): enter image description here

For every user in user_id, I need to use the values in the columns df2 as index of 'weeks' in df3, and multiply them to the respective values in df1.

E.g.: user 163, column measurements has value 0.0 (from df2). The look-up in df3 at week 0.0 is 2. The final value to be computed for this user/column is 2(from df1) times 2 = 4.

I need to estimate this for all users in user_id and all columns (activity, nutrition, etc.)

Any ideas?

I have been playing with .apply but I find it hard to structure the problem correctly.

Upvotes: 0

Views: 3736

Answers (1)

Jeff
Jeff

Reputation: 2228

The key, I think, is to put all this data together. You can work with it separately by iterating and going back and forth, but much easier and robust to use Pandas merge functionality, like this:

import pandas as pd

data1 = {'user_id':[163], 'measurements':[2.0]}
data2 = {'user_id':[163], 'measurements':[0.0]}
data3 = {'weeks':[0.0], 'measurements':[2.0]}

df1 = pd.DataFrame(data1)
df2 = pd.DataFrame(data2)
df3 = pd.DataFrame(data3)

df = df1.merge(df2, on='user_id', how='outer', suffixes=['_df1', '_df2'])
df = df.merge(df3, left_on='measurements_df2', right_on='weeks',
              how='outer', suffixes=['', '_df3'])
df['new_val'] = df['measurements_df1'] * df['measurements']

In [13]: df
Out[13]:
   measurements_df1  user_id  measurements_df2  measurements  weeks  new_val
0               2.0      163               0.0           2.0    0.0      4.0

In the future it's much easier if you give us a reproducible example to work with, especially if you can include errors with what you tried, but in this case I know what you mean about it being hard to figure out how to structure the question properly. I strongly suggest the book from the creator of Pandas, Wes McKinney.

Upvotes: 2

Related Questions