Make it possible to use the index as a variable about ggpy HOT 6 CLOSED

Just as a reference: currently you need to use df.reset_undex() to get the index as a normal column:

by_decade = ts.groupby(floor_decade).sum()
by_decade.index.name = 'year'
by_decade = by_decade.reset_index()

(from http://blog.yhathq.com/posts/aggregating-and-plotting-time-series-in-python.html -> cell 12/13)

from ggpy.

I would vote for implementing both of these

• if no x mapping is given, use the index as x mapping
• look into the column namess and the index name

Only question would be what happens when there are multiple indices? Just pick the first? Combine them?

from ggpy.

The "look into both columns and index names" and "index as x if no x is given" should be easy, will do on Tuesday/Wednesday if no one does it before me :-)

re multiindex: I see two ways for using a index when __index__ (or no x variable) is given as mapping: either combine all index into one variable (there should be a way in pandas to combine multiple columns into one... there usually is :-) or simple give an error with the hint how to combine the index into one level or only use one level of the index (by naming it directly or using reset_index). I would vote for the second way, as I suspect the "combine" step will result in strings (datetime + different firms?) and that is usually not helpfull for plotting. And automatically figuring out that this should result in a facet plot or something like this is too much magic for my taste.

from ggpy.

Has there been any movement on this?

Agree re: multi-indexing. ggplot shouldn't try to be too smart :)

from ggpy.

Currently this got "harder", as currently we let patsy handle this. One way to fix this would be to first reset the index...

# in ggplot/ggplot.py
def _apply_transforms(data, aes):
[...]
-     data = data.copy()
+     data = data.copy().reset_index()

from ggpy.

Now in #285, closing here

from ggpy.