[SciPy-user] scikit.timeseries plotting bugs?

Pierre GM pgmdevlist@gmail....
Thu Dec 4 17:29:24 CST 2008

OK, all is well...

On Dec 4, 2008, at 11:10 AM, Dharhas Pothina wrote:

> If I use two commands and '.', 'b+' or as the symbol for the first  
> series it works, ie :
> fsp1.tsplot(fseries, '.')
> fsp1.tsplot(fseries_freq, 'r--')

Yep. In that case, by the time you try to plot fseries_freq, fsp1.freq  
has been initialized to series.freq, and fseries_freq is automatically  
adjusted to match fsp1.freq

> If I use two commands and certain other symbols like 'b',b--' etc  
> for the first series, it only plots the second series :
> fsp1.tsplot(fseries, 'b.')
> fsp1.tsplot(fseries_freq, 'r--')

??? can't reproduce this one...

> If I combine the two in one command as shown in the moving average  
> plotting example in the documentation it only shows the second  
> series whatever I put as the symbol in the first series. ie :
> fsp2 = fig2.add_tsplot(111)
> fsp2.tsplot(fseries, '.',fseries_freq, 'r--')

Well, it shows the first series, not the second one... And yes, tht  
can be characterized as a bug. See, when you plot everythng with the  
same command, fsp2.freq hasn't be initialized yet. The initialization  
takes place at the end of the loop on the arguments. By then,  
fseries_freq has been plotted, but at its own frequency, viz using a  
range of xs far smaller than the fseries. So in fact, fseries-freq is  
plotted, but at the wrong xs, and you can't see it. We should update  
the frequency in real time, not at the end of the loop. Hop, something  
else on my todo list

> Am I missing something or is this a bug. Also the plots are fairly  
> slow, is that just the overhead of plotting through the timeseries  
> package?

yes, all the fancy automatic adjustment of the ticks to match your  
resolution comes to a price. And it's expensive...

More information about the SciPy-user mailing list