CODEWITHSUNDEEP
pythonpandasdataframe
TRex
TRex

Reputation: 495

how to get a specific column sequence

this is my df

df = pd.DataFrame(np.random.randint(1,10,(5,6)),columns=['a','b','c','d','e','f'])

I have to perform quite a few operations and finally I might end up with a df with column sequence changed or one or more of the columns missing

i.e. case1 - the final df column sequence is ['e','b','d','f'] or case2 - the final df column sequence is ['d','a','f','e'] or case3 - the final df column sequence is ['d','f','e','b']

how do I make sure regardless of how many columns I am left with, the final sequence of columns is [...,'d','e','f']

i.e. if am left with ['e','b','d','f'] the final df columns sequence is ['b','d','e','f']

for ['d','a','f','e'] the rearrange df columns sequence should be ['a','d','e','f']

for['c','a','b','e','f','d'] the rearrange df columns sequence should be ['a','b','c','d','e','f']

i.e. ['other columns', 'd','e','f']

my original df has more than 80 odd columns so I need to be able to do it dynamically.

Upvotes: 0

Views: 160

Answers (1)

jezrael
jezrael

Reputation: 862611

If always all values from L are in Dataframe columns use Index.difference with join list L and select by subset:

L = ['d','e','f']

df = df[df.columns.difference(L).tolist() + L]
#if order is important
#df = df[df.columns.difference(L, sort=False).tolist() + L]
print (df)
   a  b  c  d  e  f
0  8  1  4  6  2  5
1  8  7  7  8  5  9
2  8  6  2  6  5  5
3  4  9  9  5  1  5
4  2  4  2  1  1  9

If some value missing from L add Index.intersection:

df = pd.DataFrame(np.random.randint(1,10,(5,6)),columns=['a','b','c','d','e','g'])

L = ['d','e','f']

df = df[df.columns.difference(L).tolist() + df.columns.intersection(L).tolist()]
print (df)
   a  b  c  g  d  e
0  2  3  2  7  4  4
1  9  3  4  6  9  7
2  9  6  1  9  7  7
3  4  6  1  8  2  8
4  4  2  4  2  6  8

Or if need all columns from list use DataFrame.reindex:

df = df.reindex(df.columns.difference(L).tolist() + L, axis=1)
print (df)
   a  b  c  g  d  e   f
0  1  7  7  8  1  2 NaN
1  9  8  6  3  2  7 NaN
2  8  1  7  6  8  2 NaN
3  3  3  2  5  2  2 NaN
4  8  4  4  1  4  1 NaN

Upvotes: 2

Related Questions