hey everyone! sharing this for people interested in working with real-world datasets and exploring what can be done beyond basic data analysis.
historical datasets can tell us a lot about past trends, but using that information to predict future demand is a different challenge.
retail forecasting is a good example. you have historical sales, changing demand patterns and different products, and need to build models that can produce useful forecasts.
Packt is hosting a two-day hands-on Time Series Bootcamp on November 28–29 with Graeme Davidson, Lei Ma and Jeffrey Tackes, where participants work with practical retail demand forecasting.
some of the things covered:
preparing and working with time series data in Python
building forecasting models using MLForecast and StatsForecast
exploring NeuralForecast, N-BEATS, N-HiTS and TFT
experimenting with foundation models like Chronos-2 and TimeCopilot
backtesting and measuring forecasting performance
handling uncertainty with conformal methods
forecast…