#!/usr/bin/env python
# coding: utf-8

# To drop all empty columns (but still keeping the headers) using the Python [Pandas library](http://pandas.pydata.org/) we can use the following 4-line script to read in the csv file, drop the columns where **all** the elements are missing, and save the data to a new csv file.

# In[1]:


from pandas.io.parsers import read_csv
data = read_csv('remove-empty-columns.csv')
filtered_data = data.dropna(axis='columns', how='all')
filtered_data.to_csv('empty-columns-removed.csv', index=False)


# As shown below, the sample data included in the csv file has 3 columns which contain missing values.
# 
# The second column, labeled **bar**, is completely empty except the header; columns like this should be dropped. The other columns contain data, but should not be dropped even though they contain some missing values.

# In[2]:


data


# Using the [pandas.DataFrame.dropna()](http://pandas.pydata.org/pandas-docs/stable/generated/pandas.DataFrame.dropna.html) function with the **columns** axis we can drop any column where **all** the entries are **NaN** (missing values).

# In[3]:


filtered_data = data.dropna(axis='columns', how='all')
filtered_data