DOC: GH2072 add example and warning on fallback indexing for float indexes.

[adamgreenhall]

#2072 (comment)

[jreback]

can you make the doc lines wrap, no more than 80 char width. (I get flagged for this all the time!)

Looks good otherwise

@y-p ?

[jreback]

also, maybe add an example of using the 'float' selection, e.g

say A is your previous indexer column
something like this?

df[(df.A>=1.0) & (df.A<2.0)]

[ghost]

instead use use - needs fixing.

very bad idea is not the best pharsing IMO. explain the problem. explain the caveats,
recommend a better way.

Also, I can't parse:
If you do have a float index, an integer label will fallback to
it seems disjointed.

I haven't followed this issue, so i'm looking at the doc with fresh eyes... ;)

[ghost]

Oh, re 80 char limit on documentation - stop the PEP8 insanity! it's not even code!

[adamgreenhall]

Good point @y-p. How about this?

[jreback]

looks good to me (and now you are taking style points from @y-p )....!

[ghost]

Hey- if you don't want pedantry, don't ask for my opinion on docs.

[jreback]

and if you wouldn't mind, pls rebase into 1 commit

git rebase -i <first commit>

change the pick to fixup (or squash if you want) for all but the first

then

git push <your fork> <this branch name> -f

[adamgreenhall]

@jreback - done. I wondered how that worked in git - thanks.

[jreback]

this looks great

I'll leave open for a couple of days in case you want to change anything
(I find that I read doc updates and then I edit them the next day)

[ghost]

yep, this is good.

[jreback]

@adamgreenhall
in http://pandas.pydata.org/pandas-docs/dev/indexing.html#fallback-indexing

I realized that the last part of your example is correct, but not illustrating what I think you want.

df_new.index is the index in this case, but you want to illustrate df_new['index']

In [14]: df_new
Out[14]: 
   index         A         B         C         D
0      1 -1.164567  0.054407  0.619462  1.936210
1      2  0.047871 -1.776328 -0.590250 -0.399167
2      3  0.214890 -0.593469 -1.115490  0.259086
3      4  0.089140 -0.998506  0.514325 -0.719256

In [15]: df_new[df_new.index==1.0]
Out[15]: 
   index         A         B        C         D
1      2  0.047871 -1.776328 -0.59025 -0.399167

In [16]: df_new[df_new['index']==1.0]
Out[16]: 
   index         A         B         C        D
0      1 -1.164567  0.054407  0.619462  1.93621

Same here

In [17]: df_new[(df_new['index']>=1.0) & (df_new['index']<2)]
Out[17]: 
   index         A         B         C        D
0      1 -1.164567  0.054407  0.619462  1.93621

Doing something like this might make the example more clear (then you can use df_new.my_float_index
that's my 2c

df_new = df.reset_index().rename(columns = dict(index = 'my_float_index'))