I have a DataFramedf = pd.DataFrame(np.random.randint(-100, 100, 100).reshape(10, -1), columns=list(range(10)), index=list('abcdefghij')) and I found some indexes using method .idxmin(axis=1) and getting Seriesdf_ind =
- a 5
- b 8
- c 2
- d 0
- e 7
- f 0
- g 4
- h 6
- i 5
- j 2
How to get column of their values and than add (insert) it to df?
df.insert(11, "values", df[df_ind]) gives me a 10×10 DataFrame, but not 1×10 DataFrame
>Solution :
If you want to slice the values, you can use a modified indexing lookup:
# first reindex the indexer to ensure
# having the correct final order
idx, cols = pd.factorize(df_ind.reindex(df.index))
# then perform indexing lookup
df['values'] = df.reindex(columns=cols).to_numpy()[np.arange(len(df)), idx]
Output:
0 1 2 3 4 5 6 7 8 9 values
a 72 -53 17 92 -33 95 3 -91 -79 -64 95
b -13 -30 -12 40 -42 93 -61 -13 74 -12 74
c -19 65 -75 -23 -28 -91 48 15 97 -21 -75
d 75 92 -18 -1 77 -71 47 47 42 67 75
e -68 93 -91 85 27 -68 -69 51 63 14 51
f 83 -72 -66 28 28 64 -47 33 -62 -83 83
g -21 32 5 -58 86 -69 20 -99 -35 69 86
h -43 -65 2 19 -89 74 -18 -9 28 42 -18
i -1 -47 40 21 70 -16 -32 -94 96 -53 -16
j 27 31 0 80 -22 43 48 86 -77 41 0
Output if the original DataFrame was in a different order:
0 1 2 3 4 5 6 7 8 9 values
c -19 65 -75 -23 -28 -91 48 15 97 -21 -75
e -68 93 -91 85 27 -68 -69 51 63 14 51
h -43 -65 2 19 -89 74 -18 -9 28 42 -18
d 75 92 -18 -1 77 -71 47 47 42 67 75
g -21 32 5 -58 86 -69 20 -99 -35 69 86
i -1 -47 40 21 70 -16 -32 -94 96 -53 -16
b -13 -30 -12 40 -42 93 -61 -13 74 -12 74
a 72 -53 17 92 -33 95 3 -91 -79 -64 95
j 27 31 0 80 -22 43 48 86 -77 41 0
f 83 -72 -66 28 28 64 -47 33 -62 -83 83
Reproducible df_ind:
df_ind = pd.Series({'a': 5, 'b': 8, 'c': 2, 'd': 0, 'e': 7, 'f': 0, 'g': 4, 'h': 6, 'i': 5, 'j': 2})