Plotting based on groupby

Question

I have a df like so:

Year    Grass       Crop        Forest      Ecoregion   CRP
1993    30.41268857 68.45446692 0.255632102 46e         0508common
2001    47.29988968 47.68577796 0.509939614 46e         0508common
2006    71.37357063 20.40485399 0.908684114 46e         0508common
1993    27.17246635 71.97582809 0.12611897  46k         0508common
2001    65.74087991 30.61323084 0.1229253   46k         0508common
2006    81.763099   12.4386173  0.180860941 46k         0508common
1993    30.83567893 68.14034747 0.737649228 46e         05f08
2001    59.45355722 35.68378142 0.354265748 46e         05f08
2006    64.98592643 28.61787829 0.339706881 46e         05f08
1993    28.38187702 71.40776699 0.080906149 46k         05f08
2001    81.90938511 15.4368932  0.118662352 46k         05f08
2006    86.3214671  9.207119741 0.172599784 46k         05f08
1993    18.46387279 80.77686402 0.270081631 46e         05f97
2001    41.23923454 53.1703113  0.605111585 46e         05f97
2006    65.30004066 25.45626696 0.989918731 46e         05f97
1993    20.34764075 78.68863002 0.218653535 46k         05f97
2001    55.42761042 39.96085063 0.191151874 46k         05f97
2006    76.34526161 16.53176535 0.246221691 46k         05f97

and I want to create graphs based a groupby on Ecoregion. Then within each Ecoregion I want to graph based on unique CRP So each unique Ecoregion will get its own pdf file and then within that file will be graphs based on CRP. In this case Ecoregion 46e will have three graphs (0508common, 05f08 and 05f97) and Ecoregion 46k will also have three graphs.

I am trying the following code:

import pandas as pd
import matplotlib.pyplot as plt
from matplotlib.backends.backend_pdf import PdfPages
import os

df=pd.read_csv(r'C:\pathway_to_file.csv')
group=df.groupby(['Ecoregion'])
pdf_files = {}
out=r'C:\output_location'
for ecoregion, outer_group  in df.groupby(['Ecoregion']):
    with PdfPages(os.path.join(out,ecoregion + '.pdf')) as pdf:
                for crp, inner_group in outer_group.groupby(['CRP']):
                    title=crp + '_' + ecoregion
                    lu_colors=(['g','y','b','r', 'k'])
                    plot=group.plot(x=['Year'], y=['Grass', 'Crop', 'Forest'],kind='bar', colors=lu_colors, title=title).get_figure()
                    plt.legend(loc='center left', bbox_to_anchor=(1.0, 0.5))
                    plt.xticks(rotation=70)
                    plt.set_xlabel('Year')
                    plt.set_ylabel('Percent')
                    pdf.savefig(plot)  
                    plt.close(plot)

but this doesn't work properly, the graphs aren't even bar graphs like I want them to be.

An example of how to get one individual graph to be like I want it to is with this, but this doesn't use the groupby like I want it to:

with PdfPages(r'G:\graphs.pdf') as pdf: 
        lu_colors=(['g','y','b','r', 'k'])
        ax=df.set_index('Year').plot(title='0508common_46e', kind='bar', colors=lu_colors)
        plt.legend(loc='center left', bbox_to_anchor=(1.0, 0.5))
        plt.xticks(rotation=70)
        ax.set_xlabel('Year')
        ax.set_ylabel('Percent')
        fig=plt.gcf()
        pdf.savefig(fig)
        plt.close(fig) 

in this case the df would be:

    Year    Grass       Crop        Forest      Ecoregion   CRP
    1993    30.41268857 68.45446692 0.255632102 46e         0508common
    2001    47.29988968 47.68577796 0.509939614 46e         0508common
    2006    71.37357063 20.40485399 0.908684114 46e         0508common

Answer 1

You had a mistake in plot. You have to plot inner group (igr) not outer. I had slightly changed your code to be more smoothly:

import pandas as pd
import matplotlib.pyplot as plt
from matplotlib.backends.backend_pdf import PdfPages
import os

lu_colors=(['g','y','b','r','k'])

df = pd.read_csv('1.csv', header=0, usecols = [0,1,2,3,4,5])
for ecor, ogr  in df.groupby(['Ecoregion']):
    with PdfPages(os.path.join("./pdf", ecor.strip()+'.pdf')) as pdf:
        for crp, igr in ogr.groupby(['CRP']):
            title = crp.strip() + '_' + ecor.strip()
            plot = igr.plot(x=['Year'], y=['Grass', 'Crop', 'Forest'], kind='bar', colors=lu_colors, title=title).get_figure()
            plt.legend(loc='center left', bbox_to_anchor=(1.0, 0.5))
            plt.xticks(rotation=70)
            ax = plt.gca()
            ax.set_xlabel('Year')
            ax.set_ylabel('Percent')
            pdf.savefig(plot, bbox_inches='tight', pad_inches=0)  
            plt.close(plot)

One of the results: