Pandas groupby two columns and create a plot of count totals

Question

I'm new to Pandas and I'm looking for a way to plot data that has been grouped by two columns. Here's my example:

First I group by the 'Date'(year) and 'Primary Type' column.

groups = df.groupby([df['Date'].map(lambda x: x.year), pri_type['Primary Type']])

Now from that I can get a series of basically exactly what I want to plot.

groups.size().head()

Date  Primary Type        
2001  ARSON                   1010
      ASSAULT                31384
      BATTERY                93448
      BURGLARY               26011
      CRIM SEXUAL ASSAULT     1794 
dtype: int64

But when I plot this I get a very messy plot with thousands of labels on the x axis. What I would like to get is a plot with date on the x axis and a ledgend with all the Primary Types. Something similar to this graph:

Thanks in advance!

Answer 1

What do you want to be displayed on the x axis, date? If so, you can set date as index: groups.set_index('Date')

Answer 2

The solution that I came up with is to convert the series to a data frame and use the unstack() method. Here is what I did:

# convert to a dataframe
df = groups.size().to_frame()

|       |               |  0
|------ | --------------|------
|Date   | Primary Type  |
|       | ARSON         | 1010
|       | ASSAULT       | 31384
| 2001  | BATTERY       | 93234
|       | BURGLARY      | 26031
|       | CRIM SEXUAL AS| 1723

# unstack() to pivot the data which puts it in the correct format for plot()
df.unstack(level=-1)

|            |0                    
|------------|-------|---------|-------...
|Primary Type|ARSON  |ASSAULT  |BATTERY...
|Date        |       |         |       ...
|2001        |1010.0 |31384.0  |93234.0...
|2002        |2938.0 |31993.0  |94235.0...
|2003        |955.0  |30082.0  |92834.0...

Which almost makes the graph I was after, other than the 0, but I can probably get rid of that. And as you can see it's still not very readable, but this solves my question of how to graph it.

df.unstack(level=-1).plot(kind='bar', figsize = (10,10))

Final graph