README

Analyse NHS data to answer specific questions

** UNPACKING **
tar xvf AbideFinancialChallange.tar.gz

** USAGE **
Usage: python3 main.py -c <Postcode_File_Name> -a <Address_File_Name> -p <Prescription_file_name>'
The file names should be fully qualified paths if not in current directory.

** TESTS **
Navigate to the test directory using the terminal, then issue the following command:
nosetests -v

Note: This requires "nose" to be installed, which can be done using the easy_install script as follows:
sudo easy_install nose

** ADDITIONAL DATASETS **
In order to map the pharmacies to a region, I needed another dataset, which I obtained here: https://data.gov.uk/dataset/national-statistics-postcode-lookup-uk

** DESIGN **
- I went with the Iterator pattern for the Base class, FileReader.
- It overrides the __iter__ method to perform file iteration.
- There are 3 different methods to process a file:
    -- lazy_read uses the csv.DictReader to read a file, one row at a time into a dictionary. Useful for files that contain a header.
    -- chunky_lazy_read relies on reading a file, 15000 lines at a time and yielding one row at a time.
    -- lazy_sequential_read reads the file one row at a time.
    -- The last two methods require the header to be specified explicitely.
- FileReader is an abstract class, each subclass needs to implement two methods: write_output_file & process_file.
- process_file method is the workhorse. All computations are called from within here.
- All output formatting is placed in write_output_file.

** ADDITIONAL QUESTION **
- I chose to figure out the total number of Anti-depressant prescriptions per region in the UK.
- The antidepressant prescriptions I looked for are outlined here: http://www.nhs.uk/Conditions/Antidepressant-drugs/Pages/Introduction.aspx

** NOTES **
- I noticed that the two supplied datasets are from different timeframes, February 2012 for the addresses and September 2011 for the prescriptions.
- Certain prescriptions (=3269) were not mapped due to this discrepancy. I mapped these to a region "UNKNOWN". However, none of the prescriptions being scanned for fell into this category. So these are not visible in the output.