Data Science life cycle from a technical perspective
summary
Argues that data prep and clean code matter more than chasing models and shows end-to-end choices: source invoice tables (customer ID, invoice ID, amount, doc/due/payment dates, items, country, risk), derive target as payment_date − due_date with a 3‑day threshold for early vs late, run EDA with Pandas Profiling, encode/scale/aggregate features (last‑month sum, avg, country×customer), train classifiers with time‑aware splits, deploy daily Power BI dashboard via pipelines, and add monitoring scripts for automated retrain triggers.
AI-written summary · about
chapters (7)
- 0:00Introduction
- 5:22Life cycle phases
- 23:54Business understanding
- 45:40Data understanding
- 1:04:27Preprocessing & feature engineering
- 1:37:06Modeling & deployment
- 1:50:49Q&A
related talks
39mHelp Me, Help You: Supporting Your Data
39mDoing Data: The Critical Process of Data Preparation (...)
22mData Analysis in the Age of Data Democratization
31mLife cycle of data becoming a model
41mGetting Your Data AI-Ready
30mData Engineering building your BI infrastructure from scratch
33mDataEngBytes 2023 - MEL-T1-02 - Kendra Vant
40mDomain-driven Data Science