pandas rank duplicated values

Question

Lets take a dataframe of one column with random values. I want to get the rank of all these values which is easy by doing:

df.rank()

But if there are duplicated values you will get a duplicated value also for the rank. For example, for a given list of numbers:

[127.0, 131.856, 132.88, 126.249, 128.417, 124.336, 131.856, 130.624, 147.906, 134.412, 130.735, 133.433, nan, 125.59, 130.211, 133.847, 137.431, 130.0, 127.4, 132.226, 138.134]

the output of the rank function will be:

[4.0, 11.5, 14.0, 3.0, 6.0, 1.0, 11.5, 8.0, 20.0, 17.0, 9.0, 15.0, nan, 2.0, 7.0, 16.0, 18.0, 10.0, 5.0, 13.0, 19.0]

As you can see, the position 1 and 6 are the same and there is no 11 or 12 in the full list. How can we get a rank for these numbers even if it's arbitrary which one goes first?

Answer 1

Use the method parameter in rank, for example:

pd.Series(l).rank(method='first')

0      4.0
1     11.0
2     14.0
3      3.0
4      6.0
5      1.0
6     12.0
7      9.0
8     20.0
9     17.0
10    10.0
11    15.0
12     2.0
13     8.0
14    16.0
15    18.0
16     7.0
17     5.0
18    13.0
19    19.0
dtype: float64