customer-lifetime-value

KO | EN

gitlite — search

#typescript #ai-agents #ai #dsh-plugin #deepseek-harness #open-source #cli #claude-code #codex #developer-tools #react #windows

customer-lifetime-value

★ 3 Open GitHub ↗

A pyspark ETL example using a jupyter/pyspark-notebook Docker container

// repository documentation

Was this content helpful?

★ 0 (0 ratings)

## PySpark example
### ETL for Metrics Surrounding Customer Lifetime Value from Photo Sharing Event Data

The present `/src` directory includes a data directory that contains python files to create fake event data.  The python module `Faker==0.7.7` was used to create fake data as describe in the file `/src/data/fake_events.py`.  The file `/src/data/run_fake_events.py` creates a random amount within specified limits of fake events.  Global variable `DAYS=14`,  `N_CUSTOMERS=50`, and `MAX_EVENTS=5` were implemented to keep the data manageable.  The timespan over which the fake events where generated was limited to the most recent 14 days (where an event date may end up being given a week numbers from 0 to 2 depending on the definition of a week), max events per customer were set randomly between 1 to 5, and the maximum number of customers in the fake event population were 50. The fake event data was populated as json strings to the text file at `/input/input.txt`.
To generate the data the following command was run from the `/src` directory:

$ python data/run_fake_events.py

The figure below depicts the event data generation.
  
<img src="./images/event-data.png" height=50%, width=50%>

All PySpark ETL was conducted inside an ipython Jupyter notebook running inside the `jupyter/pyspark-notebook` Docker container previously created and found on DockerHub at `https://hub.docker.com/r/jupyter/pyspark-notebook`

The script used to start the jupyter/pyspark-notebook container is located in the file `/src/start_spark_notebook.sh`.  It mounts the volume of the current working directory so that data is easily uploaded from the input data directory.
The notebook is located in the `/src/notebooks` directory.  To view and edit the notebook, the authentication key is provided at start up and appended to the local host url.  Use of this container is further documented at `https://github.com/jupyter/docker-stacks/tree/master/pyspark-notebook`.  All that is required to start it up, provided Docker is installed and running the followin from the `/src` directory:

$ ./start_spark_notebook.sh

Output was generated and populated to the `/output` directory from ETL methods inside the spark notebook.  Three output files were deposited there, which provide an visualization of intermediate progress towards the final output of a customer id and customer lifetime value.

All the methods and queries used in the notebook, were exported to the file `/src/methods_pyspark_etl/methods_pyspark_etl.py`.  This file is intended to be the import for testing performed with the `pytest==3.0.6` module and included in the `/src/methods_pyspark_etl/tests` directory. The `/src/methods_pyspark_etl/tests/test_pyspark_etl.py` file is representative of some of the bugs that were found during the process of creating the pyspark notebook and still a work in progress.

customer-lifetime-value

Explore Similar Repositories

Was this content helpful?

Recent Feedback

Download README