Re: [Matplotlib-users] 3d Bar Chart

[Erik W. <er...@er...>]

Hi John,

Ok- that makes sense.

While on the subject of bar charts, can you tell me what I'm doing wrong here?

def chart(request):
    from PIL import Image as PILImage
    from matplotlib.backends.backend_agg import FigureCanvasAgg as FigureCanvas
    from matplotlib.figure import Figure
    from StringIO import StringIO
    fig = Figure()
    canvas = FigureCanvas(fig)
    ax = fig.add_subplot(111)
    from pylab import *
    N = 7
    menMeans = (20, 35, 30, 35, 27, 21, 60)
    ind = arange(N)  # the x locations for the groups
    width = 0.35       # the width of the bars
    p1 = bar(ind, menMeans, width, color='b')
    #ylabel('Time')
    title('Time In Minutes - Last 7 Days')
    xticks(ind+width, ('Mon', 'Tue', 'Wed', 'Thu', 'Fri', 'Sat', '*') )
    canvas.draw()
    size = canvas.get_renderer().get_canvas_width_height()
    buf=canvas.tostring_rgb()
    im=PILImage.fromstring('RGB', size, buf, 'raw', 'RGB', 0, 1)
    imdata=StringIO()  #ERROR LINE - 'module' object is not callable
    im.save(imdata, format='PNG')
    response = HttpResponse(imdata.getvalue(), mimetype='image/png')
    return response

I keep getting an error saying 'module' object is not callable
pointing to line 45 [imdata=StringIO()]

This method worked fine for a pie chart.

Thanks!
Erik

On 6/4/07, John Hunter <jd...@gm...> wrote:
> On 6/4/07, Erik Wickstrom <er...@er...> wrote:
>
> > I'm trying to create a 3d bar chart like the one attached for use in a
> > django site.  I have no idea where to start.  Is it even possible with
> > matplotlib?
>
> Not really, and this is by design  I've intentionally avoided all of
> these power point style features which mostly detract from the
> information in the graph.  In a bar chart, once you've displayed the
> height of the bar, there is no additional information to be had by
> adding a gradient to it, or by making it look 3D.  Those extra special
> effects confuse the information that the graph actually contains.  In
> many cases, I think that's why people want to add them: if the data
> are boring or don't say what you want, why not spice it up with some
> sexy graph features so people won't look so hard at the data?
>
> The shadow on the pie chart was a nod to those who want  "chart junk"
> as Tufte calls these effects, but it was probably a mistake.
>
>   http://www.edwardtufte.com/tufte
>
> I might accept a patch to add features like this if they remained off
> by default, but I probably won't be adding it myself.
>
> JDH
>