Why You (Currently) Do Not Need Deep Learning for Time Series Forecasting | Towards Data Science

What you need instead: Learnings from the Makridakis M5 competitions and the 2023 Kaggle AI report