I'm plotting a bunch of fault data, indexed by year and week number, and have a working bar chart.
What I want to do is instead of just plotting the count of faults each week, have a stacked bar chart that visually shows counts of generic types of faults each week (I have a column called 'Cause' for this). I think it's not working because I'm creating a new index, and don't know how to link 'Cause' counts correctly.
import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
from dateutil import rrule
from datetime import datetime
from datetime import timedelta
df = pd.read_excel(r'C:\Users\486001\OneDrive - Alstom\Documents\M33\Ramp\PLUSTWO_486001_1670756366.xlsx', sheet_name='Filtered')
df['RepDate'] = pd.to_datetime(df.reportdate, format='%b %d, %Y, %I:%M %p', errors='coerce')
df['Rep_week'] = df['RepDate'].apply(lambda x: x.isocalendar()[1])
df['Rep_year'] = df['RepDate'].apply(lambda x: x.isocalendar()[0])
df['value'] = '1'
#print(df.head(5))
week_groups = df.groupby([df['Rep_year'],df['Rep_week']])['value'].count()
start_date = datetime(2021,3,1)
end_date = datetime(2022,12,31)
flat_week_groups = week_groups.reset_index()
dummy_index = []
for dtime in rrule.rrule(rrule.WEEKLY, dtstart=start_date, until=end_date):
dt_week = dtime.isocalendar()[1]
dt_year = dtime.isocalendar()[0]
dummy_index.append(tuple([dt_year, dt_week]))
for i in dummy_index:
if i not in week_groups.index:
flat_week_groups = flat_week_groups.append(
{'Rep_year': i[0],
'Rep_week': i[1],
'value': 0},ignore_index=True)
week_groups = flat_week_groups.sort_values(['Rep_year','Rep_week']).set_index(['Rep_year','Rep_week'])
week_groups.plot(kind='bar', stacked=True, figsize=(26,7), legend=False)
I've tried groupby and including df['cause'], but this just messes up my index.
source https://stackoverflow.com/questions/74763585/stacked-bar-plot-in-python-created-index-and-mapping-to-original-dataframe
Comments
Post a Comment