5
votes

Working with Matplotlib in Python (2.7.9). I have to plot a table in a subplot (in this case subplot name is tab) but I can't seem to find a way to change the font size of the table (http://imgur.com/0Ttvzee - bottom left). Antman is happy about the results, I am not. This is the code I've been using.

EDIT: Added full code

def stat_chart(self):

DN      = self.diff
ij      = self.ij_list
mcont   = self.mcont
ocont   = self.ocont
ucont   = self.ucont
dist    = self.widths
clon    = '%1.2f' %self.mclon
clat    = '%1.2f' %self.mclat
clonlat = "{0}/{1}".format(clon,clat)
area    = self.area
perim   = self.perimeter
mdist   = np.array(self.widths)
mdist   = mdist[:,0]*10
mdist   = np.mean(mdist)
pstat   = self.polygon_status
if pstat == 1:
  status = "Overestimation"
else:
  status = "Underestimation"

# Setting up the plot (2x2) and subplots
fig   = plt.figure()
gs    = gridspec.GridSpec(2,2,width_ratios=[2,1],height_ratios=[4,1])
main  = plt.subplot(gs[0,0])
polyf = plt.subplot(gs[0,1])
tab   = plt.subplot(gs[1,0])
leg   = plt.subplot(gs[1,1])
tab.set_xticks([])
leg.set_xticks([])
tab.set_yticks([])
leg.set_yticks([])
tab.set_frame_on(False)
leg.set_frame_on(False)

# Main image on the top left
main.imshow(DN[::-1],cmap='winter')
x1,x2,y1,y2 = np.min(ij[:,1])-15,np.max(ij[:,1])+15,np.min(ij[:,0])-15,np.max(ij[:,0])+15
main.axvspan(x1,x2,ymin=1-((y1-320)/float(len(DN)-320)),ymax=1-((y2-320)/float(len(DN)-320)),color='red',alpha=0.3)
main.axis([0,760,0,800])

# Polygon image on the top right
polyf.imshow(DN,cmap='winter')
polyf.axis([x1,x2,y2,y1])
polyf.plot(mcont[:,1],mcont[:,0],'ro',markersize=4)
polyf.plot(ocont[:,1],ocont[:,0],'yo',markersize=4)
polyf.plot(ucont[:,1],ucont[:,0],'go',markersize=4)
for n,en in enumerate(dist):
  polyf.plot([en[2],en[4]],[en[1],en[3]],color='grey',alpha=0.3)

# Legend on the bottom right
mc = mlines.Line2D([],[],color='red',marker='o')
oc = mlines.Line2D([],[],color='yellow',marker='o')
uc = mlines.Line2D([],[],color='green',marker='o')
ed = mlines.Line2D([],[],color='black',alpha=0.5)
pos_p = mpatches.Patch(color='lightgreen')
neg_p = mpatches.Patch(color='royalblue')
leg.legend([mc,oc,uc,ed,pos_p,neg_p],("Model Cont.","Osisaf Cont.","Unknown Cont.","Dist. Mdl to Osi", \
  'Model Overestimate','Model Underestimate'),loc='center')

# Statistics table on the bottom left
stats = [[clonlat+' degrees' ,'%1.4E km^2' %area,'%1.4E km' %perim,'%1.4f km' %mdist,status]]
columns = ('Center Lon/Lat','Area','Perimeter','Mean Width','Status')
rows = ['TODOpolyname']
cwid = [0.1,0.1,0.1,0.1,0.1,0.1]
the_table = tab.table(cellText=stats,colWidths=cwid,rowLabels=rows,colLabels=columns,loc='center')
table_props = the_table.properties()
table_cells = table_props['child_artists']
for cell in table_cells: cell.set_height(0.5)
plt.show()

return

EDIT2: Eventually (un)solved plotting text instead of table. Good enough.

2
If you do a lot of this kind of thing, you may want to consider exporting your data to shapefile (via eg, networkx) and creating your plots in QGIS. The QGIS print composer really puts matplotlib to shame when it comes to managing multiple maps and tables. - mmdanziger
Thanks but I'm not a good programmer (as you can see from the shameful script) and I'm running out of time. Learning QGIS now and transferring everything would take too much time. For the future though it's a great tool. Thanks. - GDino

2 Answers

14
votes

I had a similar issue in changing the fontsize. Try the following

the_table.auto_set_font_size(False)
the_table.set_fontsize(5.5)

Worked for me.

2
votes

According to the docs, table has a kwarg called fontsize, a float value for the size in points.

In your example from above, for a fontsize of 5 points you would use:

    the_table =tab.table(cellText=stats,colWidths=cwid,rowLabels=rows,colLabels=columns,loc='center',fontsize=5)

If you require greater control, you can pass a FontManager instance to the cell.set_text_props() method as described in this example. That would enable you to set the family, spacing, style etc, in addition to the size.

EDIT: Playing around with Matplotlib's example, it seems that just passing fontsize to the table has no effect. However, importing

    from matplotlib.font_manager import FontProperties

and then looping through the cells and running

    cell.set_text_props(fontproperties=FontProperties(size = 5))

does have the desired effect. It is unclear why the documented kwarg fontsize does not work in this (or apparently in your) case.