How I learned Feature Store
About me
I'm a machine learning engineer has hands on experience of NLP, CV and bioinformatics.
Why I wanted to learn Feature Store
When company grow, these's tremendous data needs to be integrated and insight needs to be share. A warehouse framework which can leverage shared pipeline to save computation power and speed up the ETL process would be a huge plus for a company.
How I approached learning Feature Store
Alibaba wrote a book about feature store
Challenges I faced
Some description is too vague to comprehend.
Key takeaways
- Use multiple-layer architecture to share common data between different business unit
- Split table according to service
- Join all the fact table with dimension table at low level to prevent wasting time on join when you're doing ETL job.
medium: https://medium.com/@davidtnfsh/%E5%A4%A7%E6%95%B0%E6%8D%AE%E4%B9%8B%E8%B7%AF-%E9%98%BF%E9%87%8C%E5%B7%B4%E5%B7%B4%E5%A4%A7%E6%95%B0%E6%8D%AE%E5%AE%9E%E8%B7%B5-%E8%AE%80%E6%9B%B8%E5%BF%83%E5%BE%97-54e795c2b8c
Tips and advice
Reading this book would be highly recommended.
If not, here's my medium link about this book but in mandarin: https://medium.com/@davidtnfsh/%E5%A4%A7%E6%95%B0%E6%8D%AE%E4%B9%8B%E8%B7%AF-%E9%98%BF%E9%87%8C%E5%B7%B4%E5%B7%B4%E5%A4%A7%E6%95%B0%E6%8D%AE%E5%AE%9E%E8%B7%B5-%E8%AE%80%E6%9B%B8%E5%BF%83%E5%BE%97-54e795c2b8c
Final thoughts and next steps
Really need to discuss about this topic with other data engineer.