CODEWITHSUNDEEP
pythonpandas
Александр Сотников
Александр Сотников

Reputation: 13

How to sort values in pandas DataFrame based on a few columns without column prority?

I need to sort pandas df by values without priority of particular column. I've attached the image with expected result. In this case, index '2' has 0,99 value, so we put it at the first place. Then, we put '4' at the second row because it contains 0,95 value(independently of exact column) etc. How can I apply such kind of sorting?

enter image description here

Upvotes: 1

Views: 223

Answers (1)

Mustafa Aydın
Mustafa Aydın

Reputation: 18315

You can get the maximum values of each row, then get the indices that would sort them in descending order with np.argsort and then index into your dataframe with those:

sorted_df = df.iloc[np.argsort(df.max(axis=1))[::-1]]

where np.argsort gives the indices that would sort the maximum values of each row but in ascending order; so we use [::-1] to reverse that to get descending order. iloc performs the integer indexing,

to get

# sample dataframe
np.random.seed(51)
df = pd.DataFrame({"a": np.random.random(size=5),
                   "b": np.random.random(size=5),
                   "c": np.random.random(size=5)})
>>> df

          a         b         c
0  0.675731  0.949338  0.174343
1  0.044712  0.157670  0.218183
2  0.343304  0.387973  0.646746
3  0.644020  0.589994  0.249831
4  0.284213  0.487796  0.410248

>>> sorted_df

          a         b         c
0  0.675731  0.949338  0.174343
2  0.343304  0.387973  0.646746
3  0.644020  0.589994  0.249831
4  0.284213  0.487796  0.410248
1  0.044712  0.157670  0.218183

Upvotes: 1

Related Questions