How to *not* display 'NaN' in ipython notebook (html table of pandas dataframe)?

Question

I have a small CSV file of real-world data from tests performed on different days, etc. Not all of the same parameters were tested in each session, so there are a bunch of blank cells sprinkled around the original spreadsheet.

Tuner Location,200,210,220,230,240,250,260,270,280
07/17 #1,,,0.319,0.319,0.233,,0.215,,0.3355
07/21 #1,,0.539,0.482,0.034,0.343,0.478,0.285,0.01,0.538
07/21 #2,,,0.107,0.407,0.559,,0.185,0.439,0.36
07/21 #3,,,0.127,,,,,,
07/22 #1,0.316,0.201,0.646,,,,,,
07/22 #2,,0.098,0.138,0.134,0.194,,,,
07/22 #3,,0.216,0.187,,,,,,
07/27 #1,,0.118,0.065,0.013,1.013,,,,
08/05 #1,,,,,,,0.032,,
08/05 #2,,,,,,,0.128,,
08/05 #3,,,,,,0.235,0.159,0.324,
08/05 #4,,,,,,,0.398,,
08/05 #5,,,,,,0.214,0.121,0.121,

I'm trying to learn to manipulate and display this data in ipython notebook like I would in a regular spreadsheet program. so when I run the following lines inside a notebook:

import pandas as pd

# Set print option so the dataframe will be represented as HTML instead of plain text
pd.core.format.set_printoptions(notebook_repr_html=True)

# Read in csv file as a pandas dataframe
df = pd.read_csv('tuner-data.csv')

# View the HTML representation
df

I get a very nice looking HTML table of the data... with 'NaN' everywhere there was a blank cell in the original CSV file.

I understand 'why' NaN is necessary for later calculations, but it really makes the table hard for viewers to read (my opinion).

Is there a good/easy/simple way to suppress the display of 'NaN' in the HTML table displayed in ipython notebook?

Answer 1

This page has some suggestions. For example, you might try:

df.fillna(0)

Or:

df.fillna("")

Answer 2

A potential workaround would be to use the styles, and then show the styled output rather than the df, as the df.style.format has a na_rep parameter

    s = df.style.format(na_rep='')
    s

The advantage of using the style option is that you do not change the dataframe and therefore will not cause issues with future computations.