[FIXED] How to split a dataframe based on similarity of rows?

Issue

I have the following dataset:

            0      1       2       3
xxx_qsaqx   2      2.2    2.3      2.4
xxx_gtygs   3      3.2    3.3      3.4
xxx_uytgo   4      4.2    4.3      4.4
xxx_ytghr   5      5.2    5.3      5.4
xxy_uyhga   6      6.2    6.3      6.4
xxy-uytei   7      7.2    7.3      7.4
xyy_uiyta   8      8.2    8.3      8.4

And I want to split it into the following 3 dataframes:

xxx-df

            0      1       2       3
xxx_qsaqx   2      2.2    2.3      2.4
xxx_gtygs   3      3.2    3.3      3.4
xxx_uytgo   4      4.2    4.3      4.4
xxx_ytghr   5      5.2    5.3      5.4

xxy-df

            0      1       2       3
xxy_uyhga   6      6.2    6.3      6.4
xxy-uytei   7      7.2    7.3      7.4

xyy-df

            0      1       2       3
xyy_uiyta   8      8.2    8.3      8.4

Note that strings are indeed the row indices while 0, 1, 2, and 3 are the columns of the dataframes.

Solution

Don't try to generate variables dynamically, this is considered bad practice.

Instead, collect the DataFrames in a dictionary using a combination of str.extract to get the prefix and groupby to split the groups:

grouper = df.index.str.extract('^([^-_]+)', expand=False)

dfs = {k: g for k,g in df.groupby(grouper)}

Output:

{'xxx':            0    1    2    3
        xxx_qsaqx  2  2.2  2.3  2.4
        xxx_gtygs  3  3.2  3.3  3.4
        xxx_uytgo  4  4.2  4.3  4.4
        xxx_ytghr  5  5.2  5.3  5.4,
 'xxy':            0    1    2    3
        xxy_uyhga  6  6.2  6.3  6.4
        xxy-uytei  7  7.2  7.3  7.4,
 'xyy':            0    1    2    3
        xyy_uiyta  8  8.2  8.3  8.4,
}

And access your sub-DataFrames with:

dfs['xxx']

           0    1    2    3
xxx_qsaqx  2  2.2  2.3  2.4
xxx_gtygs  3  3.2  3.3  3.4
xxx_uytgo  4  4.2  4.3  4.4
xxx_ytghr  5  5.2  5.3  5.4

Answered By - mozway