How to get maximum value from dataframe with numeric indexes and headings

Question

I have the following dataframe:

df:

Unnamed: 0          0          1
0          0.0   0.000000        NaN
1          1.0   2.236068   0.000000
2          2.0   3.000000   2.236068
3          NaN   0.000000   1.000000
4          0.0   0.000000        NaN
5          1.0   1.414214   0.000000
6          2.0   2.828427   1.414214
7          NaN   0.000000   1.000000
8          0.0   0.000000        NaN
9          1.0   3.162278   0.000000
10         2.0  11.401754   3.162278
11         NaN   0.000000   1.000000
12         0.0   0.000000        NaN
13         1.0  14.142136   0.000000
14         2.0   2.828427  14.142136

I'm trying to get the maximum value from each set of data, the problem is that I generated this dataframe from several operations, but at the end the indexes and headings are numbers and I can not use groupby or loc. What I need is something as follows:

df1
        0
 1   3.000000
 2   2.828427
 3  11.401754
 4  14.142136

Answer 1

You absolutely use loc ! Problem is that you aren't paying attention to whether df.columns are integers or strings. Since you're having issues, I'm guessing strings.

However, what you're trying to do is not at all clear

IIUC

m = df['0'] == 0
g = m.cumsum()[~m]
df.loc[~m, '0'].groupby(g).max()

Answer 2

Use .iloc and cumsum :

df.groupby((~df.iloc[:,0].astype(bool)).cumsum()).max()

Output:

            Unnamed: 0          0          1
Unnamed: 0                                  
1                  2.0   3.000000   2.236068
2                  2.0   2.828427   1.414214
3                  2.0  11.401754   3.162278
4                  2.0  14.142136  14.142136

To just get the maxes for column index 1:

df.groupby((~df.iloc[:,0].astype(bool)).cumsum()).max().iloc[:,1]

Output:

Unnamed: 0
1     3.000000
2     2.828427
3    11.401754
4    14.142136
Name: 0, dtype: float64

Answer 3

file.csv :

0,1,2,3
9,6,7,
0,,,
5,6,2

Try:

import pandas as pd

df = pd.read_csv('file.csv', header=-1)
# keep only max per row
print(df.max(axis=1))

Output:

0    3.0
1    9.0
2    0.0
3    6.0
dtype: float64