Time Series with Deep Learning (LSTM, TFT, N-BEATS), GenAI (Amazon Chronos), Prophet, Silverkite, ARIMA. Demand Forecast
InstructorDiogo Alves de Resende
Editorial course preview
These 3 complementary views highlight concrete, legible examples from the course presentation.
This image displays a Jupyter Notebook environment where Python code is used to load a dataset from a CSV file, set the date as the index, and prepare the data for time series analysis.
The notebook interface shows the implementation of an LSTM model using a custom RNN module, highlighting hyperparameters such as hidden dimensions and dropout rates.
The notebook shows Python code using pandas resample methods to plot quarterly seasonality and applying seasonal_decompose to analyze Bitcoin demand patterns.
Instruction spans traditional Exponential Smoothing and ARIMA to modern architectures like Temporal Fusion Transformers (TFT), N-BEATS, and GenAI via Amazon Chronos.
The curriculum utilizes multiple capstone projects, such as Air Miles and Bike Sharing, to apply specific forecasting techniques to real-world datasets.
One signal suggests a lack of explicit statistical testing or detailed mathematical explanations for fundamental neural network components. Note that this review predates the displayed update, and the update label does not prove these gaps were addressed.
Lessons may include significant time spent debugging or looking up commands, which can interrupt the learning momentum. This signal also predates the displayed update; the update label does not guarantee these pacing issues were resolved.
The curriculum aligns with the target audience of analysts and data scientists through its Python-centric approach to forecasting.