CODEWITHSUNDEEP
pythonpandas
leorar
leorar

Reputation: 63

How to Replace Values in Pandas Rows with Ranks

I have a dataframe with multiple columns, each of which has their own value.

data = {'name' : ['bill', 'joe', 'steve'],
    'test1' : [7, 75, 85],
    'test2' : [35, 45, 83],
     'test3' : [51, 61, 45]}
df = pd.DataFrame(data)

    name  test1  test2  test3
0   bill      7     35     51
1    joe     75     45     61
2  steve     85     83     45

I would like to replace the values in some of the columns with relative rank across the row rather than actual value. The output would be as follows.

    name  test1  test2  test3
0   bill      3     2      1
1    joe      1     3      2
2  steve      1     2      3

Is there a way to do this?

Upvotes: 3

Views: 1597

Answers (3)

piRSquared
piRSquared

Reputation: 294278

Mind-Bending Numpy Assignment

a = np.argsort(df.iloc[:, 1:].to_numpy(), axis=1)
n, m = a.shape
b = np.empty_like(a)
c, d = np.mgrid[:n, :m]
b[c, a] = m - d

df.iloc[:, 1:] = b

df

    name  test1  test2  test3
0   bill      3      2      1
1    joe      1      3      2
2  steve      1      2      3

Upvotes: 3

orlp
orlp

Reputation: 117691

>>> df.rank("columns", ascending=False)
   test1  test2  test3
0    3.0    2.0    1.0
1    1.0    3.0    2.0
2    1.0    2.0    3.0

>>> rankcols = ["test1", "test2", "test3"]
>>> df[rankcols] = df[rankcols].rank("columns", ascending=False).astype(int)
>>> df
    name  test1  test2  test3
0   bill      3      2      1
1    joe      1      3      2
2  steve      1      2      3

Upvotes: 2

Vaishali
Vaishali

Reputation: 38415

You can use DataFrame.rank on axis 1

df = df.assign(**df.iloc[:, 1:].rank(axis = 1, ascending = False).astype(int))

    name    test1   test2   test3
0   bill    3       2       1
1   joe     1       3       2
2   steve   1       2       3

Upvotes: 7

Related Questions