Row based style.apply doesn't modify background colors

[shri]

Code Sample, a copy-pastable example if possible

# add color gradient to column(s)
import numpy as np
import math
from pandas import *
from IPython.display import display, HTML

# color gradient
COLORS = ["#F1F2B5", "#CCD7A5", "#A7BC96", "#82A186", "#5D8677", "#386B67", "#135058"]

# rows/columns to skip applying gradient to
SKIPROWSORCOLUMNS = ["Day","Users"] 

# index in row or columns to skip
SKIPINDICES = [0,1]

# dataframe to operate on
df = DataFrame([["Monday",100,.65,.39,.33],["Tuesday",120,.45,.32,.31],["Wednesday",150,.75,.33,.30]])
df.columns = ["Day","Users","M1","M2","M3"] #column headers

# takes dataframe and colors
# returns colormap
def applyColorGradient(data, colors=COLORS, skiprowsorcolumns=SKIPROWSORCOLUMNS, skipindices=SKIPINDICES):
  numberOfColors = len(colors)
  i = 0
  maxNumberUnset = True
  maxNumber = 0
  while (i < len(data)):
    if (i not in skipindices):
      if(maxNumberUnset):
        maxNumber = data[i]
        maxNumberUnset = False
      else:
        current = data[i]
        if(current > maxNumber):
          maxNumber = current
    i = i + 1
  name = data.name
  dontskip = data.name not in skiprowsorcolumns
  colorArray = [colors[int(math.floor(value/maxNumber*(numberOfColors-1)))] if dontskip and index not in skipindices else '' for index,value in enumerate(data)]
  print(colorArray)
  colorSeries = pandas.Series(["background-color:" + color if color else '' for color in colorArray], name=data.name)
  display(colorSeries)
  return colorSeries


# applying colorgradient to dataframe
gradientdf = df.style.apply(applyColorGradient, axis=1) #0 is used for columns

# display table
display(gradientdf) #0 is used for columns

Expected Output

With axis = 0, this manages to give each cell a background color that is indicated. When I change this to a row based function, this doesn't color any of the cells, even though the color series are coming in. Is this a bug or am I doing something wrong?

output of pd.show_versions()

INSTALLED VERSIONS

commit: None
python: 3.5.1.final.0
python-bits: 64
OS: Linux
OS-release: 3.13.0-91-generic
machine: x86_64
processor:
byteorder: little
LC_ALL: en_US.UTF-8
LANG: en_US.UTF-8

pandas: 0.17.1
nose: None
pip: 8.1.2
setuptools: 23.0.0
Cython: 0.23.5
numpy: 1.10.4
scipy: 0.17.1
statsmodels: 0.6.1
IPython: 4.2.0
sphinx: None
patsy: 0.4.1
dateutil: 2.5.2
pytz: 2016.3
blosc: None
bottleneck: None
tables: None
numexpr: 2.5.2
matplotlib: 1.5.1
openpyxl: None
xlrd: None
xlwt: None
xlsxwriter: None
lxml: None
bs4: None
html5lib: None
httplib2: None
apiclient: None
sqlalchemy: 1.0.13
pymysql: None
psycopg2: None
Jinja2: None

[shri]

It looks like removing the pandas.Series fixed things--is pandas.Series expected to work here?

[TomAugspurger]

It's a bit hard to say, but I think your example will work if you remove the Series construction entirely.

colorSeries = pandas.Series([...])  # just return the list here

The key part is you need the output labels to match the input labels. Series.apply, which is used inside Styler.apply automatically handles the labels for you, which is why this isn't documented explicitly, but perhaps we can add a note to the documentation.

[attack68]

the heart of this issue is the following code lines:

if axis is not None:
    result = data.apply(func, axis=axis, result_type="expand", **kwargs)
    result.columns = data.columns

The DataFrame.apply method is asymmetric in how it deals with broadcasting labels to the index or the columns:

>>> df = pd.DataFrame([[1,2], [3,4]], index=["a", "b"], columns=["x", "y"])

>>> df.apply(lambda s: ["one", "two"], axis=0)
     x    y
a  one  one
b  two  two

>>> df.apply(lambda s: pd.Series(["one", "two"], index=["c", "d"]), axis=0)
     x    y
c  one  one
d  two  two

Along axis=0, the native index is maintained unless overwritten by the function result object.

>>> df.apply(lambda s: ["one", "two"], axis=1, result_type="expand")
     0    1
a  one  one
b  two  two

>>> df.apply(lambda s: pd.Series(["one", "two"], index=["c", "d"]), axis=1, result_type="expand")
     c    d
a  one  one
b  two  two

Along axis=1 (with result_type=expand), the native index is overwritten by RangeIndex unless overwritten by the function result object.

Therefore the Styler code manually updates the column names of the result to be equivalent to incoming data.

The alternative of using result_type=broadcast does nothing other than generate problems and not solve the issue.