Query regarding design of a new API (GSoC 2016)

Hi Developers,

I'm Yeshwanth from IIITDM Jabalpur, India. I want to develop a new API for
matplotlib which can automatically plot most of the visualizations required
by data scientists. For example plots like the one shown in this
stackoverflow issue (
http://stackoverflow.com/questions/14391959/heatmap-in-matplotlib-with-pcolor)
can be developed with a single function using the new API.

The API can deal with certain types of data as mentioned here
https://github.com/numfocus/gsoc/blob/master/2016/ideas-list-matplotlib.md#categorical-axes
.

Please let me know the things I've to consider before designing this API.

regards,
Yeshwanth
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/matplotlib-devel/attachments/20160305/2b2428d9/attachment.html>

Yeshwanth,

Have you seen the seaborn package?
http://stanford.edu/~mwaskom/software/seaborn/

It builds off of matplotlib to make lots of nice statistical plots.
-paul

···

On Fri, Mar 4, 2016 at 11:43 PM, Yeshwanth Sai Charan < yeshwanthsaicharan at gmail.com> wrote:

Hi Developers,

I'm Yeshwanth from IIITDM Jabalpur, India. I want to develop a new API
for matplotlib which can automatically plot most of the visualizations
required by data scientists. For example plots like the one shown in this
stackoverflow issue (
python - Heatmap in matplotlib with pcolor? - Stack Overflow)
can be developed with a single function using the new API.

The API can deal with certain types of data as mentioned here
https://github.com/numfocus/gsoc/blob/master/2016/ideas-list-matplotlib.md#categorical-axes
.

Please let me know the things I've to consider before designing this API.

regards,
Yeshwanth

_______________________________________________
Matplotlib-devel mailing list
Matplotlib-devel at python.org
Matplotlib-devel Info Page

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/matplotlib-devel/attachments/20160309/9634e5fa/attachment.html&gt;

You might also consider contributing to yhat's ggplot project
http://ggplot.yhathq.com/ or at least looking at R's ggplot2 for
inspiration.

Gary

···

On 10 March 2016 at 12:01, Paul Hobson <pmhobson at gmail.com> wrote:

Yeshwanth,

Have you seen the seaborn package?
seaborn: statistical data visualization — seaborn 0.13.0 documentation

It builds off of matplotlib to make lots of nice statistical plots.
-paul

On Fri, Mar 4, 2016 at 11:43 PM, Yeshwanth Sai Charan < > yeshwanthsaicharan at gmail.com> wrote:

Hi Developers,

I'm Yeshwanth from IIITDM Jabalpur, India. I want to develop a new API
for matplotlib which can automatically plot most of the visualizations
required by data scientists. For example plots like the one shown in this
stackoverflow issue (
python - Heatmap in matplotlib with pcolor? - Stack Overflow)
can be developed with a single function using the new API.

The API can deal with certain types of data as mentioned here
https://github.com/numfocus/gsoc/blob/master/2016/ideas-list-matplotlib.md#categorical-axes
.

Please let me know the things I've to consider before designing this API.

regards,
Yeshwanth

_______________________________________________
Matplotlib-devel mailing list
Matplotlib-devel at python.org
Matplotlib-devel Info Page

_______________________________________________
Matplotlib-devel mailing list
Matplotlib-devel at python.org
Matplotlib-devel Info Page

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/matplotlib-devel/attachments/20160310/ce30d902/attachment.html&gt;

I would like to pull in the scope of this a bit (which was a bit unclear in
the initial posting). The bulk of the work will not be in a 'charting'
API, but on making sure that the mpl internals will easily support such an
interface. As others have pointed out there are already pointed out there
are several high-level packages that are built on top of mpl and we do not
want to pull the domain-specific aspects into core matplotlib.

For example, one thing we should support is to configure an `Axes` such that

fig, ax = plt.subplots()
ax.SOMEMETHOD(??)
ax.plot(['a', 'b', 'c'], [1, 2, 3])
ax.plot(['a', 'c'], [4, 5])
ax.bar(['a', 'b', 'c'], [.5, .75, 1])

'does the right thing' to first order.

Providing this functionality at the core will take a lot of burden off of
ex seaborn and pandas plotting.

Tom

···

On Wed, Mar 9, 2016 at 9:47 PM gary ruben <gary.ruben at gmail.com> wrote:

You might also consider contributing to yhat's ggplot project
http://ggplot.yhathq.com/ or at least looking at R's ggplot2 for
inspiration.

Gary

On 10 March 2016 at 12:01, Paul Hobson <pmhobson at gmail.com> wrote:

Yeshwanth,

Have you seen the seaborn package?
seaborn: statistical data visualization — seaborn 0.13.0 documentation

It builds off of matplotlib to make lots of nice statistical plots.
-paul

On Fri, Mar 4, 2016 at 11:43 PM, Yeshwanth Sai Charan < >> yeshwanthsaicharan at gmail.com> wrote:

Hi Developers,

I'm Yeshwanth from IIITDM Jabalpur, India. I want to develop a new API
for matplotlib which can automatically plot most of the visualizations
required by data scientists. For example plots like the one shown in this
stackoverflow issue (
python - Heatmap in matplotlib with pcolor? - Stack Overflow)
can be developed with a single function using the new API.

The API can deal with certain types of data as mentioned here
https://github.com/numfocus/gsoc/blob/master/2016/ideas-list-matplotlib.md#categorical-axes
.

Please let me know the things I've to consider before designing this API.

regards,
Yeshwanth

_______________________________________________
Matplotlib-devel mailing list
Matplotlib-devel at python.org
Matplotlib-devel Info Page

_______________________________________________
Matplotlib-devel mailing list
Matplotlib-devel at python.org
Matplotlib-devel Info Page

_______________________________________________
Matplotlib-devel mailing list
Matplotlib-devel at python.org
Matplotlib-devel Info Page

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/matplotlib-devel/attachments/20160312/861b357b/attachment.html&gt;