Create automatic bar chart subplots in Python-CodePudding

I have a dataframe like so:

Date        ID  Data
2022-05-31  21  1
2022-05-31  22  4
2022-05-31  28  3
2022-06-01  21  2
2022-06-01  22  9
2022-06-01  28  2
2022-06-02  22  12
2022-06-03  22  10
2022-06-03  23  1
2022-07-03  25  1
2022-07-04  22  12
2022-07-04  23  6
2022-07-04  25  7
2022-08-05  22  12
2022-08-05  23  12
2022-08-05  25  11
2022-08-06  22  8
2022-08-06  23  11
2022-08-06  25  11

Here the x-axis is Date and y-axis is Data. The data is grouped by ID which would be the legend. I am able to create a stacked bar chart with ID as the color. However I would like create separate bar charts for each ID. In the example there are 5 unique IDs. So there would be a matrix of 5 bar charts.

Is there a way to do this automatically without splitting the dataset into groups and then plotting each ID individually?

CodePudding user response：

You can use groupby and plot the groups individually:

for key, grp in df.groupby('ID'):
    grp.plot.bar('Date', 'Data', title=key, legend=False)

For a fancier look with shared axes you can use a seaborn

CodePudding user response：

Since you had plotly in the tags, plotly.express has the ability to automatically create faceted graphs. faceted graphs by ID. If you prefer two columns, set face_col_wrap=2. Also, the color bar is displayed, so I hide it. I also exclude unnecessary dates. I create an index of minimum to maximum dates and then create an exclusion list by deleting dates where data exists.

df.sort_values(['ID','Date'], ascending=[True,True], inplace=True)

import plotly.express as px
import pandas as pd

fig = px.bar(df, x='Date', y='Data', color='ID', facet_col='ID', facet_col_wrap=1)

fig.update_layout(autosize=False,
                  height=600,
                  width=800,
                  coloraxis_showscale=False
                 )

fig.update_xaxes(
    tickvals=df['Date'].unique(),
    rangebreaks=[
        dict(values=pd.date_range(df['Date'].min(), df['Date'].max()).drop(df['Date'].unique()), dvalue=86400000)
    ])

fig.show()