Merge on single level of MultiIndex

Question

Is there any way to merge on a single level of a MultiIndex without resetting the index?

I have a “static” table of time-invariant values, indexed by an ObjectID, and I have a “dynamic” table of time-varying fields, indexed by ObjectID+Date. I’d like to join these tables together.

Right now, the best I can think of is:

dynamic.reset_index().merge(static, left_on=['ObjectID'], right_index=True)

However, the dynamic table is very big, and I don’t want to have to muck around with its index in order to combine the values.

Asked By: Johann Hibschman

||

Source

Answer 1

I get around this by reindexing the dataframe merging to have the full multiindex so that a left join is possible.

# Create the left data frame
import pandas as pd
idx = pd.MultiIndex(levels=[['a','b'],['c','d']],labels=[[0,0,1,1],[0,1,0,1]], names=['lvl1','lvl2'])
df = pd.DataFrame([1,2,3,4],index=idx,columns=['data'])

#Create the factor to join to the data 'left data frame'
newFactor = pd.DataFrame(['fact:'+str(x) for x in df.index.levels[0]], index=df.index.levels[0], columns=['newFactor'])

Do the join on the subindex by reindexing the newFactor dataframe to contain the index of the left data frame

df.join(newFactor.reindex(df.index,level=0))

Answered By: closedloop

Answer 2

Yes, since pandas 0.14.0, it is now possible to merge a singly-indexed DataFrame with a level of a multi-indexed DataFrame using .join.

df1.join(df2, how='inner') # how='outer' keeps all records from both data frames

The 0.14 pandas docs describes this as equivalent but more memory efficient and faster than:

merge(df1.reset_index(),
      df2.reset_index(),
      on=['index1'],
      how='inner'
     ).set_index(['index1','index2'])

The docs also mention that .join can not be used to merge two multiindexed DataFrames on a single level and from the GitHub tracker discussion for the previous issue, it seems like this might not of priority to implement:

so I merged in the single join, see #6363; along with some docs on
how to do a multi-multi join. That’s fairly complicated to actually
implement. and IMHO not worth the effort as it really doesn’t change
the memory usage/speed that much at all.

However, there is a GitHub conversation regarding this, where there has been some recent development https://github.com/pydata/pandas/issues/6360. It is also possible achieve this by resetting the indices as mentioned earlier and described in the docs as well.

Update for pandas >= 0.24.0

It is now possible to merge multiindexed data frames with each other. As per the release notes:

index_left = pd.MultiIndex.from_tuples([('K0', 'X0'), ('K0', 'X1'),
                                        ('K1', 'X2')],
                                        names=['key', 'X'])

left = pd.DataFrame({'A': ['A0', 'A1', 'A2'],
                     'B': ['B0', 'B1', 'B2']}, index=index_left)

index_right = pd.MultiIndex.from_tuples([('K0', 'Y0'), ('K1', 'Y1'),
                                        ('K2', 'Y2'), ('K2', 'Y3')],
                                        names=['key', 'Y'])

right = pd.DataFrame({'C': ['C0', 'C1', 'C2', 'C3'],
                      'D': ['D0', 'D1', 'D2', 'D3']}, index=index_right)

left.join(right)

Out:

            A   B   C   D
key X  Y                 
K0  X0 Y0  A0  B0  C0  D0
    X1 Y0  A1  B1  C0  D0
K1  X2 Y1  A2  B2  C1  D1

[3 rows x 4 columns]

Answered By: joelostblom

Answer 3

I would use mapping for a single column:

df1['newcol'] = df1.index.get_level_values(-1).map(lambda x: df2.newcol[x])

Answered By: Andor

Answer 4

This works for me!

gData.columns = gData.columns.droplevel(0)

grpData = gData.reset_index()
grpData

pd.merge(grpData,cusData,how='inner')

Here gData is multi index dataframe with two levels and cusData is a single index dataframe.

Answered By: Muthu

Merge on single level of MultiIndex

Question:

Answers:

Update for pandas >= 0.24.0