Pandas loop over groupby and plot each group

Question

I am trying to loop over a groupby object and plot each group. But I am having some issues. Can someone please tell me where I am going wrong?

df = pd.DataFrame([['item1',2000,1, 2], ['item1',2001,1, 2], ['item1',2002,1, 2], 
              ['item2',2000,1, 2], ['item2',2001,1, 2], ['item2',2002,1, 2]],
              columns=['mykey', 'year','val1','val2'])

grouped = df.groupby('mykey')
for name,group in grouped:
  fig = plt.figure()
  ax1 = fig.add_subplot(111)
  group.val1.plot.line(ax=ax1, ylim=[5,20], color='red',x=group.year)
  ax1.set_ylabel('val1')
  ax2 = ax1.twinx()
  group.val2.plot.line(ax=ax2, ylim=[5,20], color='blue' ,x=group.year)
  ax2.set_ylabel('val2')
  plt.title(str(name), fontsize=15);

It seems I'm close, but just there are some issues some where.

First issue is that there are 5 groups in the groupby object. I get the 5 figures as I want, but only first one has the plots(lines) on it. Others figures are blank with the correct title on them, any idea what is wrong with my code?
How can I set a group column / key as the x axis, I have tried this x=group.desiredx but it doesn't do anything.

mykey| year| val1| val2
item1| 2000| 5| 34
item2| 2001| 45| 34
item3| 2002| 34| 34
item1| 2000| 22| 65
item2| 2001| 34| 54
item3| 2002| 12| 54
item1| 2000| 23| 54
item2| 2001| 34| 34
item3| 2002| 21| 21

Please read and understand [mcve]. The question on [How to make good reproducible pandas examples](https://stackoverflow.com/questions/20109391/how-to-make-good-reproducible-pandas-examples) may help you to produce one such example. — ImportanceOfBeingErnest, Nov 29 '17 at 23:43
Thanks Brad, I have added some data. This is what the dataframe looks like before groupby. — SANM2009, Nov 29 '17 at 23:47
sorry, desiredx is year. I was making the code generic when I posted it. — SANM2009, Nov 29 '17 at 23:58

ImportanceOfBeingErnest · Accepted Answer · 2017-11-30T00:26:44.320

1

The values to plot on the y axis are 1 and 2. You set the ylim to something bigger than 5, ylim=[5,20]. Hence you do not see the values at 1 and 2.

Setting the ylim to some reasonable numbers, e.g. ylim=[0,3] will allow you to see the data.

Also, group.val1.plot will plot the data against its index, because there is no "year" in group.val1. Instead take "val1" as the y value.

import pandas as pd
import matplotlib.pyplot as plt

df = pd.DataFrame([['item1',2000,1, 2], ['item1',2001,1, 2], ['item1',2002,1, 2], 
              ['item2',2000,1, 2], ['item2',2001,1, 2], ['item2',2002,1, 2]],
              columns=['mykey', 'year','val1','val2'])

grouped = df.groupby('mykey')
for name,group in grouped:
  fig = plt.figure()
  ax1 = fig.add_subplot(111)
  group.plot.line(ax=ax1, ylim=[0,3], color='red',x="year",y="val1")
  ax1.set_ylabel('val1')
  ax2 = ax1.twinx()
  group.plot.line(ax=ax2, ylim=[0,3], color='blue',x="year",y="val2")
  ax2.set_ylabel('val2')
  plt.title(str(name), fontsize=15)

plt.show()

edited Nov 30 '17 at 00:26

answered Nov 30 '17 at 00:05

ImportanceOfBeingErnest

321,279
53
665
712

Thanks, The x is still not working. I get 0, 0.25,0.50 ..... on the x axis. The plots are y axis plots are now working. – SANM2009 Nov 30 '17 at 00:11
The year is there now, but with a trailing .00 , .25 , .50 ....... And each year is repeated 3 times. so: 2000.00 , 2000.25 , 2000.50, 2001.00 and so on – SANM2009 Nov 30 '17 at 00:32
Yes, that is to be expected. You will find a lot of questions and answers on formatting x axis labels around. – ImportanceOfBeingErnest Nov 30 '17 at 00:37
Wow this is so close. I have been working on this for days. Can you suggest a solution for the year format issue? I am searching for options too. – SANM2009 Nov 30 '17 at 00:44
Google for "matplotlib format xaxis whole numbers" or "matplotlib format dates as years" or similar. – ImportanceOfBeingErnest Nov 30 '17 at 00:46
Fixed it by adding : xticks = [2000,2001,2002] – SANM2009 Nov 30 '17 at 00:47

score 0 · Answer 2 · answered Nov 30 '17 at 00:56

Thanks to ImportanceOfBeingErnest, I have got the result what I wanted. Here is full tweaked version of the code, for anyone who may need it in future. I have added legend = False, without this val1 and val2 are written on top of each other and looks messy.

import pandas as pd
import matplotlib.pyplot as plt

 df = pd.DataFrame([['item1',2000,1, 2], ['item1',2001,1, 2], ['item1',2002,1, 2], 
          ['item2',2000,1, 2], ['item2',2001,3, 2], ['item2',2002,1, 2]],
          columns=['mykey', 'year','val1','val2'])

 grouped = df.groupby('mykey')
 for name,group in grouped:
  fig = plt.figure()
   ax1 = fig.add_subplot(111)
  group.plot.line(ax=ax1, ylim=[0,3], color='red',x="year",y="val1", legend 
  = False, xticks = [2000,2001,2002])
  ax1.set_ylabel('val1  ssssssss')
  ax2 = ax1.twinx()
  group.plot.line(ax=ax2, ylim=[0,3], color='blue',x="year",y="val2", legend 
  = False, xticks = [2000,2001,2002])
  ax2.set_ylabel('val2 dddddd')
  plt.title(str(name), fontsize=15)

   plt.show()

Pandas loop over groupby and plot each group

2 Answers2

Linked