跨列计数值，然后按年份分组 pandas

Question

1   2   3   4   year
a   h   f   h   2000
r   r   f   h   2000
h   y   g   h   2001
h   i   g   e   2004
g   f   g   b   2006
g   d   g   v   2006

Is there a way in pandas to sum the frequency of each value by year?

I tried stack(), and groupby(), but that didn't work. I'm not sure what is the next thing to try. I don't think it's suited to cross tab.

Answer 1

Use DataFrame.melt for unpivot with GroupBy.size :

df = df.melt(id_vars='year').groupby(['year','value']).size().reset_index(name='count')
print (df)
    year value  count
0   2000     a      1
1   2000     f      2
2   2000     h      3
3   2000     r      2
4   2001     g      1
5   2001     h      2
6   2001     y      1
7   2004     e      1
8   2004     g      1
9   2004     h      1
10  2004     i      1
11  2006     b      1
12  2006     d      1
13  2006     f      1
14  2006     g      4
15  2006     v      1

Your solution should be changed:

df1 = (df.set_index('year')
         .stack()
         .groupby(level=0)
         .value_counts()
         .rename_axis(['year','value'])
         .reset_index(name='count'))
print (df1)