带有时间戳记的SQL导出中按行业分组的按月划分的Pandas线图

Question

Newbie question, thank you in advance!

I'm trying to group the data by both date and industry and display a chart that shows the different industry revenue numbers across the time series in monthly increments.

I am working from a SQL export that has timestamps, having a bear of time getting this to work.

Posted sample csv data file here: https://drive.google.com/open?id=0B4xdnV0LFZI1WGRMN3AyU2JERVU

Here's a small data example:

Industry     Date                Revenue
Fast Food   01-05-2016 12:18:02  100
Fine Dining 01-08-2016 09:17:48  110
Carnivals   01-18-2016 10:48:52  200

My failed attempt is here:

import pandas as pd
import datetime
import matplotlib.pyplot as plt

df = pd.read_csv('2012_to_12_27_2016.csv')

df['Ship_Date'] = pd.to_datetime(df['Ship_Date'], errors = 'coerce')
df['Year'] =  df.Ship_Date.dt.year
df['Ship_Date'] =  pd.DatetimeIndex(df.Ship_Date).normalize()
df.index = df['Ship_Date']
df_skinny = df[['Shipment_Piece_Revenue', 'Industry']]

groups = df_skinny[['Shipment_Piece_Revenue', 'Industry']].groupby('Industry')
groups = groups.resample('M').sum()
groups.index = df['Ship_Date']

fig, ax = plt.subplots()
groups.plot(ax=ax, legend=False)
names = [item[0] for item in groups]
ax.legend(ax.lines, names, loc='best')

plt.show()

Answer 1

You could use DataFrame.Series.unique to get a list of all industries and then, using DataFrame.loc , define a new DataFrame object that only contains data from a single Industry.

Then if we set the Ship Date column as the index of the new DataFrame , we can use DataFrame.resample , specify the frequency as months and call sum() to get the total revenue for that month.

import pandas as pd
import matplotlib.pyplot as plt

df = pd.read_csv('Graph_Sample_Data.csv')
df['Ship Date'] = pd.to_datetime(df['Ship Date'], errors='coerce')

fig, ax = plt.subplots()

for industry in df.Industry.unique():
    industry_df = df.loc[df.Industry == industry]
    industry_df.index = industry_df['Ship Date']
    industry_df = industry_df.resample('M').sum()
    industry_df.plot(x=industry_df.index, 
                     y='Revenue', 
                     ax=ax,
                     label=industry)

plt.show()