Udacity Data Engineer Nanodegree

Data Modeling with Postgres

Introduction

A startup called Sparkify wants to analyze the data its been collecting on songs and user activity on their new music streaming app. The analytics team is particularly interested in understanding what songs users are listening to. Currently, Sparkify does not have an easy way to query its data, which resides in a directory of JSON logs of user activity on the app, as well as a directory with JSON metadata on the songs in their app.

The company would like a data engineer to create a Postgres database with tables designed to optimize queries on song play analysis.

Project Description and Purpose

Using skills learned about data modeling with Postgres and building ETL pipelines with Python, define fact and dimension tables for a star schema, and write an ETL pipeline that transfers data from files in two local directories into Postgres tables using Python and SQL.

After creating a relational database and standing up a reliable ETL pipeline to ingest additional data, Sparkify will be able to systematically store and easily access data associated with its application users and their activity. The data can then be used by the Sparkify analytics team to extract valuable insights related to application users' behaviors that can be leveraged for application enhancements such as song play recommendations.

Datasets

The song dataset is a subset of real data from the Million Song Dataset. Each file is in JSON format and contains metadata about a song and the artist of that song. The files are partitioned by the first three letters of each song's track ID. For example, here are filepaths to two files in this dataset.

song_data/A/B/C/TRABCEI128F424C983.json
song_data/A/A/B/TRAABJL12903CDCF1A.json

Here is an example of what a single song file, TRAABJL12903CDCF1A.json, looks like.

{"num_songs": 1, "artist_id": "ARJIE2Y1187B994AB7", "artist_latitude": null, "artist_longitude": null, "artist_location": "", "artist_name": "Line Renaud", "song_id": "SOUPIRU12A6D4FA1E1", "title": "Der Kleine Dompfaff", "duration": 152.92036, "year": 0}

The log dataset consists of log files in JSON format generated by this event simulator based on the songs in the dataset above. These simulate activity logs from a music streaming app based on specified configurations.

The log files in the dataset are partitioned by year and month. For example, here are filepaths to two files in this dataset.

log_data/2018/11/2018-11-12-events.json
log_data/2018/11/2018-11-13-events.json

And below is an example of what the data in a log file, 2018-11-12-events.json, looks like.

Database Schema

The database is designed using a star schema consisting of the following fact and dimension tables. The star schema is appropriate given the simplicity of the data model and the presence of one fact table with accompanying dimension tables. The star schema also supports the stated use case by the Sparkify analytics team to easily query data and quickly output aggregations.

Fact Table

songplays - records in log data associated with song plays i.e. records with page NextSong

column name	data type	condition
songplay_id	serial	primary key
start_time	timestamp	not null
user_id	int	not null
level	varchar
song_id	varchar
artist_id	varchar
session_id	int	not null
location	varchar
user_agent	varchar

Dimension Tables

users - users in the app

column name	data type	condition
user_id	int	primary key
first_name	varchar
last_name	varchar
gender	char
level	varchar

songs - songs in music database

column name	data type	condition
song_id	varchar	primary key
title	varchar	not null
artist_id	varchar	not null
year	int
duration	numeric

artists - artists in music database

column name	data type	condition
artist_id	varchar	primary key
location	varchar
latitude	float
longitude	float

time - timestamps of records in songplays broken down into specific units

column name	data type	condition
start_time	timestamp	primary key
hour	int
day	int
week	int
month	int
year	int
weekday	int

Steps

Below are the steps taken to complete the project:

1. Created Tables

• Wrote CREATE and DROP statements in sql_queries.py to create and drop each table for ease of creating and resetting database tables.
• Ran create_tables.py to create the database and tables.
• Ran test.ipynb to confirm the creation of the tables with the correct columns.

2. Built ETL Processes

• Followed the instructions in the etl.ipynb notebook to develop the skeleton ETL processes for each table.
• Ran test.ipynb to confirm records were successfully entered into each table.

3. Built ETL Pipeline

• Using etl.ipynb, coded all necessary functions in etl.py to process all datasets.
• Confirmed all records were successfully inserted into each table by querying the database.

Conclusion

By creating a Postgres database and ETL code to ingest new data, Sparkify and its analytics team now have the tools it needs to better understand how its users are interacting with its music streaming app. Because the project is Pythonic and all scripts have been intuitively structured, the Sparkify analytics team will easily be able to provide ongoing code maintenance and enhancements.

Example Queries

The number of song plays by hour to understand when during the day users are using the app more frequently - app usage appears to peak between 3-6pm.

select
t.hour,
count(distinct sp.songplay_id) as song_plays
from songplays sp
join time t on t.start_time = sp.start_time
group by t.hour
order by t.hour;

The number of song plays by subscription level - paid subscribers are more active.

select
level,
count(distinct songplay_id) as song_plays
from songplays
group by level;

The count of users by gender - there are more female users than male.

select
gender,
count(distinct user_id)
from users
group by gender;