Track

Time Series

Goal: Learn forecasting for ordered observations, from stationarity and classical baselines through deep sequence models, attention-based multi-horizon forecasters, and simple linear checks on transformer claims.

Prereqs: ML Basics. Deep Learning helps for the neural forecasting papers.

Status: done

Work through the steps in order. Bold links open YouTube.

Step	Concept	YouTube	Read
1	Why ordered data needs special care	Why Are Time Series Special?	Darts — Overview of forecasting models
2	Stationarity	Time Series Talk — Stationarity	Wikipedia — Stationary process
3	Autocorrelation and partial autocorrelation	Time Series Talk — Autocorrelation and Partial Autocorrelation	Wikipedia — Autocorrelation
4	ARIMA	Time Series Talk — ARIMA Model	statsmodels — Autoregressive Integrated Moving Average (ARIMA) Tutorial
5	Exponential smoothing	What are Exponential Smoothing Models	statsmodels — Exponential smoothing
6	Time-aware evaluation splits	Evaluating Time Series Models	scikit-learn — TimeSeriesSplit
7	LSTM sequence forecasting	Time Series Forecasting With RNN (LSTM)	Christopher Olah — Understanding LSTM Networks
8	DeepAR probabilistic forecasting	DeepAR — Probabilistic forecasting with autoregressive recurrent networks	Salinas et al. 2017 — DeepAR: Probabilistic Forecasting with Autoregressive Recurrent Networks
9	N-BEATS basis expansion	N-BEATS: Neural basis expansion analysis for interpretable time series forecasting	Oreshkin et al. 2019 — N-BEATS
10	Temporal Fusion Transformer	Temporal Fusion Transformers, EXPLAINED	Lim et al. 2019 — Temporal Fusion Transformers for Interpretable Multi-horizon Time Series Forecasting
11	Informer for long sequences	Informer: Time series Transformer — EXPLAINED	Zhou et al. 2020 — Informer: Beyond Efficient Transformer for Long Sequence Time-Series Forecasting
12	Linear baselines vs transformers	LTSF-Linear: Are Transformers Effective for Time Series Forecasting	Zeng et al. 2022 — Are Transformers Effective for Time Series Forecasting?
13	PatchTST patching	How PatchTST and Chronos differ for Time Series Forecasting	Nie et al. 2022 — A Time Series is Worth 64 Words: Long-term Forecasting with Transformers

Found a broken link or an unclear step? Report a problem with this track.