Data Science for Music Streaming

We sell our music through State51, and every month they send us an awful, hard-to-comprehend spreadsheet with our sales data in it. So I thought I'd see if I could make it more useful with some Ruby and Docker

Getting the data

Download the "detail view" spreadsheet, open it in Google Sheets and export it as a CSV, then place it in the data/ dir with a sensible ISO8801-compliant name (2019-10-08 or something)

Building the Docker image

make build

ought to do it

Running it

Populate the fields list in conf.yaml with the indeces of the fields you'd like to anaylyse (8, 9, 1 and 24, which we're using, correspond to Album Title, Track Title, Music Service and Country of Sale), then

make analyse

will attempt to parse the newest file it finds under data/ and should generate some output like:

********************************************************************************

Data from 2019-10-18.csv:

  The top 3 results for Album Title are:
    * Junk Science (146 occurences)
    * Flux (27 occurences)
    * Takatsuka (23 occurences)

  The top 10 results for Track Title are:
    * Hippy Jazz (73 occurences)
    * Hope to Go (14 occurences)
    * New York Stuntman (14 occurences)
    * Could Be (13 occurences)
    * Chemosphere (12 occurences)
    * Fabric (12 occurences)
    * Basic Era (8 occurences)
    * Boot (8 occurences)
    * The Cone Mill (6 occurences)
    * Ceramic Dragon (4 occurences)

  The top 10 results for Music Service are:
    * Spotify - Stream (36 occurences)
    * iTunes - EU Apple Music (27 occurences)
    * iTunes - JP Apple Music (27 occurences)
    * iTunes - US Apple Music (24 occurences)
    * iTunes - UK Apple Music (12 occurences)
    * iTunes - CA Apple Music (9 occurences)
    * iTunes - RoW Apple Music (6 occurences)
    * iTunes - BR Apple Music (5 occurences)
    * Google Play - Google Locker GB (5 occurences)
    * iTunes - RU Apple Music (4 occurences)

  The top 10 results for Country of Sale are:
    * GB (56 occurences)
    * JP (29 occurences)
    * US (27 occurences)
    * DE (13 occurences)
    * CA (9 occurences)
    * UA (6 occurences)
    * FR (6 occurences)
    * BR (5 occurences)
    * RU (4 occurences)
    * DK (3 occurences)

Possibly we made £4.43 in sales, but it's hard to tell

********************************************************************************

Developing it

If you'd like to play with the code (and God knows it needs help) then something like

make run

will drop you into a shell from which you can run

ruby analyse.rb /opt/data/2019-10-18.csv

or similar

Running the tests

There are no tests