Time Series Analysis with StatsModels
summary
Shows how to build time‑series models from first principles with pandas then switch to StatsModels: compute a 12‑month moving average, extract a repeated monthly seasonal_decompose pattern, fit additive vs multiplicative model, and build AR/MA/I(S) predictors and ARIMA-style forecasts. Uses EIA monthly renewable electricity (2001–present) and Our World in Data surface temperatures (1950–present). Seasonal decomposition captures most variance (R² ≈ 0.905 for electricity); solar projections consistently underpredict actual rapid growth.
AI-written summary · about
chapters (28)
- 0:00Time series intro
- 5:44Seasonal decomposition basics
- 13:13Moving average trends
- 16:48Detrending and seasonal extraction
- 23:25Residual analysis and R-squared
- 28:35Seasonal decomposition recap
- 30:01Temperature data exercise
- 34:04Statsmodels auto decomposition
- 37:26Fetching temperature data
- 41:57Temperature data decomposition
- 46:58Multiplicative models intro
- 48:21Solar power setup
- 50:11Train-test split
- 51:23Multiplicative decomposition
- 54:00Quadratic trend modeling
- 57:33Plotting predictions
- 1:00:53Prediction accuracy
- 1:03:07Q&A
- 1:06:17ARIMA models intro
- 1:10:41Year-over-year differences
- 1:11:47Auto regression with lags
- 1:15:00AR model fit
- 1:17:31ARIMA moving average
- 1:19:36Combining AR and MA
- 1:21:01Reconstructing values
- 1:23:54ARIMA components
- 1:26:28Statsmodels ARIMA
- 1:27:10Summary and resources
related talks
01h 43m(Serious) Time for Time Series
29mA dive into time series for the energy sector
01h 18m(Serious) Time for Time Series
29mTime Series Forecasting Techniques in Python
01h 40mApplied Time Series Econometrics in Python and R
01h 26mApplying Statistical Modeling & Machine Learning to Perform Time-Series Forecasting
26mTime Series Analysis in Python: Easy Tools for Scientific Insight
44mMethods and Tools for Time Series Data Science Problems