Reputation: 753
I have a DataFrame with a MultiIndex fopr the columns.
ipdb> actions
flow inflow outflow
action Investment Trade ExternalFee Fee
date sequence
2016-10-18 50 15000.0 NaN NaN NaN
55 NaN NaN -513.0 NaN
60 NaN -14402.4 NaN NaN
70 NaN NaN NaN -14.29
I wish to reindex thereby adding and 'Income' column.
ipdb> actions.reindex(columns=['Investment', 'Trade', 'ExternalFee', 'Fee', 'Income'], level=1)
flow inflow outflow
action Investment Trade ExternalFee Fee
date sequence
2016-10-18 50 15000.0 NaN NaN NaN
55 NaN NaN -513.0 NaN
60 NaN -14402.4 NaN NaN
70 NaN NaN NaN -14.29
No 'Income' column is added.
I also tried naming the level:
ipdb> actions.reindex(columns=['Investment', 'Trade', 'Income'], level='action')
flow inflow outflow
action Investment Trade
date sequence
2016-10-18 50 15000.0 NaN
55 NaN NaN
60 NaN -14402.4
Upvotes: 2
Views: 1317
Reputation: 863801
You need reindex
by all columns - so need export MultiIndex
to tuples, add value and last reindex:
tuples = actions.columns.tolist()
tuples = tuples + [('outflow','Income')]
print (tuples)
[('inflow', 'Investment'), ('outflow', 'Trade'),
('outflow', 'ExternalFee'), ('outflow', 'Fee'),
('outflow', 'Income')]
a = actions.reindex(columns=pd.MultiIndex.from_tuples(tuples))
print (a)
inflow outflow
Investment Trade ExternalFee Fee Income
2016-10-18 50 15000.0 NaN NaN NaN NaN
55 NaN NaN -513.0 NaN NaN
60 NaN -14402.4 NaN NaN NaN
70 NaN NaN NaN -14.29 NaN
Another posible solution is:
actions[('outflow','Income')] = np.nan
print (actions)
action inflow outflow
date Investment Trade ExternalFee Fee Income
2016-10-18 50 15000.0 NaN NaN NaN NaN
55 NaN NaN -513.0 NaN NaN
60 NaN -14402.4 NaN NaN NaN
70 NaN NaN NaN -14.29 NaN
Upvotes: 2