3

Help please,

A, B and C are columns

X , Y , Z, Q , R are my index values

I want to make a stacked bar plot in python (matplot lib) which calculates the normalized - percentage values for each OUT_CAUSE per market and represents it in the chart.

 OUT_CAUSE  A   B   C   
 MARKET                                             
 X  12.0    11.0    4.0
 Y  9.0     12.0    3.0
 Z  16.0    40.0    1.0
 Q  23.0    19.0    3.0
 R  17.0    15.0    2.0

X axis of the plot should be MARKET Y axis of the plot should be % contribution of OUT_CAUSE for each market.

Sum of the % should be equal to 100

I was using this for representation of direct numbers

df(kind='bar', stacked=True ,figsize=(8, 8));

pratyada
  • 49
  • 4

2 Answers2

2

You can have a loop to create %age for all columns: 

for colname in df.columns: 
    colsum = sum(df[colname])
    df[colname+'%'] = round(100*df[colname]/colsum,1)  # create new column for %ages

Output:

      A     B    C    A%    B%    C%
X  12.0  11.0  4.0  15.6  11.3  30.8
Y   9.0  12.0  3.0  11.7  12.4  23.1
Z  16.0  40.0  1.0  20.8  41.2   7.7
Q  23.0  19.0  3.0  29.9  19.6  23.1
R  17.0  15.0  2.0  22.1  15.5  15.4

You can then use following command to create your stacked bar chart: 

df.iloc[:,3:].T.plot.bar(stacked=True, rot=0)
plt.show()

enter image description here

If you cannot count the number of columns, you can filter out columns names by presence of % character: 

colnames = list(filter(lambda x: '%' in x, df.columns.tolist()))
df[colnames].T.plot.bar(stacked=True, rot=0)
plt.show()
rnso
  • 23,686
  • 25
  • 112
  • 234
1

A simple (but probably not the most effective) way to do it is to simply create a new column:

df['percentage'] = df['A']/df['A'].sum()

And then you simply plot the 'percentage' column.

NotAName
  • 3,821
  • 2
  • 29
  • 44