How to add a colorbar legend next to a calmap.calendarplot?

[BreitA]

Hi,

I'm currently heavily relying on your module to plot calendar heatmap. Your module is great but I wonder If you know a simple way to put a colorbar for the whole figure on the left of the calendar heatmap figure. It could also be a neat feature in your next iteration on the module to have an option to put automatically the colorbar when creating the heatmap.

[martijnvermaat]

Hi @BreitA,

Yes, this is something that I would like to have. Currently, I don't think it's even possible to do this manually outside of calmap after calling it.

An implementation in yearplot would probably start with something like this:

diff --git a/calmap/__init__.py b/calmap/__init__.py
index 1c53a8c..fddca78 100644
--- a/calmap/__init__.py
+++ b/calmap/__init__.py
@@ -181,7 +181,8 @@ def yearplot(data, year=None, how='sum', vmin=None, vmax=None, cmap='Reds',
     # Draw heatmap.
     kwargs['linewidth'] = linewidth
     kwargs['edgecolors'] = linecolor
-    ax.pcolormesh(plot_data, vmin=vmin, vmax=vmax, cmap=cmap, **kwargs)
+    mesh = ax.pcolormesh(plot_data, vmin=vmin, vmax=vmax, cmap=cmap, **kwargs)
+    ax.figure.colorbar(mesh)

     # Limit heatmap to our data.
     ax.set(xlim=(0, plot_data.shape[1]), ylim=(0, plot_data.shape[0]))

Some quick thoughts:

1. This would automatically add another axis, which I'm not sure I like from an API point of view (currently, yearplot simply works on one axis, which it returns).
2. I sense it will be difficult to get the sizing right, especially in the easy "just works" use case.
3. Needs additional customization options (e.g., via a cbar_kws argument, and by providing an existing axis).
4. Perhaps this works better on calendarplot than on yearplot?

Not sure when I'll have time to look into this. Would welcome a PR though :)

[BreitA]

I cooked something this evening that works, I will send you what I've done when it's clean.

[BreitA]

Okay so it's very "homemade" but does the job. I'm very bad with matplotlib so I'm sure there is a more elegant way to do the stuff but it works wonders for me atm. as you can see I didt it as an external function to call after the calmap plot. The function add another axis cb_ax that is the colorbar to the figure. This is the copy of my notebook EDIT : update 2016-03-10 a little debug going through after testing it at work + most options tested + a little bit of pep8 ISSUES :

• when the calendar plots autoscale to fit the figure the colorbar doesn't autoscale too (see last example)

def add_colorbar(data, fig, cmap, auto_tick=True,
                 vmin=None,
                 vmax=None,
                 num_label=3,
                 cb_pos=None,
                 decimal=0,
                 yticks=None,
                 ytick_labels=None,
                 cbar_width=0.01):
    # colorbar position
    if cb_pos is None:
        axes = fig.get_axes()
        if len(axes) > 1:
            pos1 = axes[0].get_position()
            pos2 = axes[-1].get_position()

            cb_pos = [pos1.x1 + 0.05, pos2.y0, cbar_width, pos1.y1 - pos2.y0]
        else:
            cb_pos = [0.9 + 0.05, 0.4, cbar_width, 0.22]  # fine custom pos for the single year plot object

    # custom generation of colorbar (seen in http://matplotlib.org/examples/color/colormaps_reference.html) 
    gradient = np.linspace(0, 255, 256)
    gradient = np.vstack((gradient, gradient))
    # generate colorbar ax
    cb_ax = fig.add_axes(cb_pos)
    cb_ax.imshow(gradient.T, aspect='auto', cmap=plt.get_cmap(cmap))

    # generate tick and tick label by hand (I'm sure there is a much more efficient thing to do here)
    if vmin is None:
        vmin = np.min(data)
    if vmax is None:
        vmax = np.max(data)
    vmax = float(vmax)
    vmin = float(vmin)
    if auto_tick is True:

        step = (vmax - vmin) / (num_label - 1)
        ytick_labels = np.arange(vmin, vmax, step)
        ytick_labels = np.append(ytick_labels, vmax)
        ytick_labels = np.round(ytick_labels, decimal)
        ytick_labels = [str(v) for v in ytick_labels.tolist()]
        yticks = tuple(np.linspace(0, 255, num_label).tolist())

    elif yticks is None or ytick_labels is None:

        raise TypeError('yticks and ytick_labels required if auto_tick is False')
    elif len(yticks) != len(ytick_labels):

        raise TypeError('yticks and ytick_labels must be of same length')
    else:
        # custom ticks
        num_label = len(yticks)
        yticks = np.asarray(yticks) - vmin
        yticks = yticks * 255 / (vmax - vmin)
    # edit ytick + remove xticks  
    cb_ax.set_yticks(yticks)
    cb_ax.set_xticks([])
    cb_ax.set_yticklabels(ytick_labels)
    cb_ax.yaxis.tick_right()
    return cb_ax

#few years example
df=pd.DataFrame(data=np.random.randn(900,1)
                ,index=pd.date_range(start='2014-01-01 00:00:00',freq='1D',periods =900)
                ,columns=['data'])
cmap='RdYlGn'
fig,ax=calmap.calendarplot(df['data'],
                    fillcolor='grey', linewidth=0,cmap=cmap,
                    fig_kws=dict(figsize=(17,8)))
fig.suptitle('Calendar view' ,fontsize=20,y =1.08)

cb_ax=add_colorbar(df['data'],fig,cmap=cmap)

cb_ax.tick_params(axis='y', which='major', labelsize=15)
cb_ax.set_title('rand_data',fontsize=15,y=1.02)      

#few years example custom
df=pd.DataFrame(data=np.random.randn(900,1)
                ,index=pd.date_range(start='2014-01-01 00:00:00',freq='1D',periods =900)
                ,columns=['data'])
cmap='RdYlGn'
yticks=[-5,1,2,4.5]
ytick_labels=['-5 git','1 git','2 git','4.5 git' ]
fig,ax=calmap.calendarplot(df['data'],
                    fillcolor='grey', linewidth=0,cmap=cmap,
                    fig_kws=dict(figsize=(17,8)),vmin=-5,vmax=5)
fig.suptitle('Calendar view' ,fontsize=20,y =1.08)

cb_ax=add_colorbar(df['data'],fig,cmap=cmap,vmin=-5,vmax=5,decimal=1,
                   yticks=yticks,
                   ytick_labels=ytick_labels,
                   auto_tick=False,
                  cbar_width=0.03)

cb_ax.tick_params(axis='y', which='major', labelsize=15)
cb_ax.set_title('rand_data',fontsize=15,y=1.02)       

# lots of years example
df=pd.DataFrame(data=np.random.randn(2000,1)
                ,index=pd.date_range(start='2014-01-01 00:00:00',freq='1D',periods =2000)
                ,columns=['data'])
cmap='RdYlGn'
fig,ax=calmap.calendarplot(df['data'],
                    fillcolor='grey', linewidth=0,cmap=cmap,
                    fig_kws=dict(figsize=(17,17)))
fig.suptitle('Calendar view' ,fontsize=20,y =1.08)

cb_ax=add_colorbar(df['data'],fig,cmap)

cb_ax.tick_params(axis='y', which='major', labelsize=15)
cb_ax.set_title('rand_data',fontsize=15,y=1.02)       

# yearplot test
df=pd.DataFrame(data=np.random.randn(365,1)
                ,index=pd.date_range(start='2014-01-01 00:00:00',freq='1D',periods =365)
                ,columns=['data'])
cmap='RdYlGn'

fig=plt.figure(figsize=(17,8))

ax=calmap.yearplot(df['data'],
                    fillcolor='grey', linewidth=0,cmap=cmap,
                    )

cb_ax=add_colorbar(df['data'],fig,cmap)
cb_ax.tick_params(axis='y', which='major', labelsize=15)
cb_ax.set_title('rand_data',fontsize=15,y=1.08)   
fig.suptitle('Calendar view' ,fontsize=20,y=0.7)

# lots of years example bad scale 
df=pd.DataFrame(data=np.random.randn(2000,1)
                ,index=pd.date_range(start='2014-01-01 00:00:00',freq='1D',periods =2000)
                ,columns=['data'])
cmap='RdYlGn'
fig,ax=calmap.calendarplot(df['data'],
                    fillcolor='grey', linewidth=0,cmap=cmap,
                    fig_kws=dict(figsize=(17,8)))
fig.suptitle('Calendar view' ,fontsize=20,y =1.08)

cb_ax=add_colorbar(df['data'],fig,cmap)

cb_ax.tick_params(axis='y', which='major', labelsize=15)
cb_ax.set_title('rand_data',fontsize=15,y=1.02)       

[martijnvermaat]

@BreitA Thanks a lot for sharing your code.

I don't have the time to work on it now, but I would like to include something like you're showing here in the future. So I'm leaving this issue open, thanks again :+1:

[kidpixo]

Hi, I was searching for the same stuff and I didn't think about looking here in the issues!

I answer to a question on stackoverflow python 2.7 - add a colorbar to a calmap plot and I think it's yours, @BreitA .

My solution is to pick the children of the axis returned by yearplot and stick the colorbar to this mappable : after digging in the code, it is the second call to ax.quadmesh . So the code would be:

fig,ax=calmap.calendarplot(df['data'],
                    fillcolor='grey', linewidth=0,cmap='RdYlGn',
                    fig_kws=dict(figsize=(17,8)))

fig.colorbar(ax[0].get_children()[1], ax=ax.ravel().tolist())

and even simpler for a single yearplot (it is encapsulated in an explicit call to figure to declare the size) :

fig = plt.figure(figsize=(20,8))
ax = fig.add_subplot(111)
cax = calmap.yearplot(df, year=2014, ax=ax, cmap='YlGn')
fig.colorbar(cax.get_children()[1], ax=cax, orientation='horizontal')

A good start would be to return the reference to the actual graphic object that contains the data in the yearplot function

[martijnvermaat]

Thanks @kidpixo for commenting here and linking back to the SO thread!

I'm interested in adding this or making it easier to do it yourself, but time limited at the moment.

[kidpixo]

Thank you for the package, @martijnvermaat !

A solution would be to alter the return of yearplot, I don't know how this impact back compatibility.

I can also think about an helper function, like yearplotcolorbar who takes the output of yearplot and wrap the code I used to generate the plot.

Something more elaborated for calendarplot should do the trick.

[BreitA]

Thanks for the solution provided @kidpixo , will try it when I have the time ! And yes the Stackoverflow was mine and then I ended up asking the question here.

[kidpixo]

@BreitA you are welcome! Could you please accept my solution on stackoverflow?

[BreitA]

Yeah I just did! I tested it and it is exactly what I was looking for and it works great!

[kidpixo]

Nice to hear :-D if you have time to contribute or to publish some code of your solution will be great!

[BreitA]

Honestly at this point your solution is much more elegant and show a better understanding of matplotlib than I have done with my homemade solution.
I will use your solution for my future use of this lib tbh.

Thanks again.

[dnaneet]

> fig = plt.figure(figsize=(20,8))
> ax = fig.add_subplot(111)
> cax = calmap.yearplot(df, year=2014, ax=ax, cmap='YlGn')
> fig.colorbar(cax.get_children()[1], ax=cax, orientation='horizontal')
> ```
> 

This works beautifully!

[MarvinT]

Hi, if this problem still exists and you'd like to create a PR to fix it please direct it to https://github.com/MarvinT/calmap/ That is the version that gets published to pypi and has received several updates to fix some existing issues.

[martijnvermaat]

Thank you for creating the issue. Unfortunately I don't have the time to maintain this project. As per @MarvinT 's comment, please see https://github.com/MarvinT/calmap/ instead.