It's Time for Py: Time Series Clustering and Anomaly Detection
summary
A client pipeline processes almost 7 million invoice records across 28 months for ~459 locations and 6,000+ products, uses dtw clust hierarchical clustering (shape and dynamic time warp distances) to reduce 459 stores to seven clusters, applies decomposition with statsmodels to separate trend, seasonality and residuals, and fits a sigmoid recovery fit to anomalies (e.g., Hurricane Harvey) to measure recovery length and slope for forecasting and operational planning.
AI-written summary · about
chapters (6)
- 0:00Intro
- 0:19Key takeaways
- 2:19Time Series Analysis
- 3:00Deep Dive: ML Pipeline
- 4:43Market CLUSTERING & segmentation
- 9:38[DRAFT] Forecasting RECOVERY TRAJECTORY
related talks
44mMethods and Tools for Time Series Data Science Problems
36mDrift Detection on Irregular Time Series with Multiple Non-Uniform Seasonal Patterns
45mAnomaly Detection in Time Series
16mDetecting Anomalies in Logs with Time Series Analysis
45mForecasting at Scale
41mApplying Foundational Models for Time Series Anomaly Detection
01h 21mBuilding a ML Model Observability pipeline
39m10 things I learned about writing data pipelines in Python and Spark.