Reformat WDI data extract for Tableau.

reshape.py

# -*- coding: utf-8 -*-

import pandas as pd

in_file_name = 'WDIData.csv'   # Make sure this file is in the same folder.

data = pd.read_csv(in_file_name)
data.replace('..', 'Null', inplace=True) # Replace WDI's ".." for missing data with "Null".
data.dropna(axis=1, how='all', inplace=True) # Drop columns with exclusively NA/Null data.
data.drop(data.columns[[1, 3]], axis=1, inplace=True) # Remove country code, series code.
data.rename(columns=lambda x: x.split(' [')[0], inplace=True) # Remove column name part after brackets. Turns "1960 [YR1960]" into "1960".

multi = data.set_index(['Series Name', 'Country Name']) # Multiindex by indicator then country.
"""
New format
Indicator  |Country    |2000   |2001 |...

GDP        |USA        |$1     |$1.50
           |Ukraine    |$1000  |$2000
Birth Rate |USA        |X1     |X2
           |Ukraine    |Y1     |Y2
etc...
"""
stacked = multi.stack() # Stack the year columns into rows indexed under indicator then country.
"""
New format
Indicator  |Country    |Year   |Value

GDP        |USA        |2000   |$1
           |           |2001   |$1.50

           |Ukraine    |2000   |$1000
           |           |2001   |$2000

Birth Rate |USA        |2000   |X1
                       |2001   |X2

           |Ukraine    |2000   |Y1
           |           |2001   |Y2
etc...
"""
unstack = stacked.unstack(level='Series Name') # Move indicators to columns.
"""
New format
Indicators:        |GDP    |Birth Rate
Country    |Year   |Value  |Value

USA        |2000   |$1     |X1
           |2001   |$1.50  |X2

Ukraine    |2000   |$1000  |Y1
           |2001   |$2000  |Y2
etc...
"""
unstack.index.names = ["Country Name", "Year"] # Rename indices.
unstack.to_csv('WDIData_reformat.csv') # Save to file.