1. Suppose I have dataframe (approx., 10 columns) focused on three columns. Let's say 'A', 'B' & 'C'.
2. 'A' column has some continuous value between any range. E.g., the price of any item is between 5-20 bucks.
3. 'B' column is categorical. E.g., it has two categories, like 'Old', 'New'
4. 'C' column is like a unique ID for that item.
5. My motive is to find the top 10 items which should be sorted by their rank in price & rank should be separated by categories mentioned in column 'B'.
6. Result is required in the plot (seaborn/matplotlib). Barplot should show top 10 IDs from column C, and each bar should have it's price from column A, this should be sorted rank-wise from higher price to lower price (plot should show bar FOR EACH CATEGORY FROM COLUMN B)

Someone please help to make related code in Python using Seaborn/Matplotlib libraries.Example table like below :

       A    B     C
0    5.0  Old  A001
1    6.2  New  A002
2   10.0  Old  A003
3   19.6  Old  A004
4   12.0  Old  A005
5   11.0  New  A006
6    7.0  New  A007
7    8.0  Old  A008
8    7.0  New  A009
9    5.0  New  A010
10  17.0  Old  A011
11   8.0  Old  A012
12  12.0  Old  A013
13  13.0  New  A014
14  15.0  New  A015
15   9.0  Old  A016
16   9.0  New  A017
17  10.0  Old  A018

CodePudding user response：

new answer top 5 per group

df2 = df.loc[df.groupby('B', group_keys=False)['A'].nlargest(5).index]
df2.set_index('C')['A'].plot.bar(color=df2['B'].map({'New': 'r', 'Old': 'b'}).values)

output:

old answer

IIUC, you want the top ten items in A and sort the output by B and A:

df2 = df.loc[df['A'].nlargest(10).index].sort_values(by=['B', 'A'])

output:

       A    B     C
5   11.0  New  A006
13  13.0  New  A014
14  15.0  New  A015
15   9.0  Old  A016
2   10.0  Old  A003
17  10.0  Old  A018
4   12.0  Old  A005
12  12.0  Old  A013
10  17.0  Old  A011
3   19.6  Old  A004

Then plot using:

df2.set_index('C')['A'].plot.bar(color=df2['B'].map({'New': 'r', 'Old': 'b'}).values)

output:

CodePudding user response：

Below is the basic code which fulfilled my requirement, however, later I used Seaborn's barplot instead.

import matplotlib.pyplot as plt
old = df[df['B']=='Old'].sort_values(by=['A'], ascending=False).head(5)
new = df[df['B']=='New'].sort_values(by=['A'], ascending=False).head(5)
fig, a = plt.subplots(1, 2, figsize=(4,4))
old.plot.bar('C', ax=a[0])
new.plot.bar('C', ax=a[1])