(Serious) Time for Time Series
summary
Shows hands‑on work with an air‑quality dataset (California, 2007–2017) using Jupyter notebooks and Pandas to parse timestamps, resample, and compute rolling means (7, 20, 50, 365‑day windows). Demonstrates rolling average smoothing to expose trend and weekly/weekday aggregations to find worst periods, then builds a forecasting baseline with linear regression with one‑hot month encoding and column transformers. Introduces gradual/gaussian seasonal filters to smooth month‑to‑month jumps and isolate trend versus seasonality for modeling and outlier detection.
AI-written summary · about
chapters (5)
- 0:00Introduction and setup
- 10:15Time series basics
- 29:00Smoothing techniques
- 43:50Trend modeling
- 58:31Seasonality and wrap-up
related talks
01h 43m(Serious) Time for Time Series
44mMethods and Tools for Time Series Data Science Problems
01h 26mApplying Statistical Modeling & Machine Learning to Perform Time-Series Forecasting
01h 30mTime Series Analysis with StatsModels
29mA dive into time series for the energy sector
01h 27mBayesian Time Series
26mTime Series Analysis in Python: Easy Tools for Scientific Insight
01h 27mIntroduction to Machine Learning with Time Series