Title here
Summary here
Module · 6 lessons
Sequence Models (RNNs & LSTMs)
Model sequential and time-series data with RNNs, LSTMs, and GRUs.
At a glance
Level
Advanced
Lessons
6 lessons
Time to complete
2–3 weeks
Cost
Free forever · no sign-up
Sequences — text, audio, stock prices — need models with memory. This module covers recurrent neural networks and their more capable LSTM and GRU variants, and applies them to real time-series forecasting.
Lessons in this module
1
Introduction to Recurrent Neural Networks
Understand why feedforward networks fail on sequences and how recurrent neural networks use a hidden state to carry memory across time steps
2
Basic RNN Architecture
Build and train a SimpleRNN in Keras to forecast the S&P 500 from sliding windows of monthly prices
3
Advanced RNN Architecture: GRU and LSTM
Understand why plain RNNs forget and how gated cells (LSTM and GRU) fix the vanishing-gradient problem, then compare all three on a real S&P 500 forecasting task
4
Adding Convolutional Layers to RNNs
Learn how to use Conv1D layers to extract local temporal patterns before an LSTM, with causal padding so the model never peeks at the future
5
Time-Series Forecasting with RNNs
Build the full time-series forecasting workflow with an LSTM in Keras: windowing, the temporal train/test split, train-only scaling, and RMSE/MAE evaluation on the S&P 500.
6
Guided Project: Forecasting the S&P 500
Assemble an end-to-end LSTM forecasting pipeline on real S&P 500 data, then evaluate it honestly and discuss its limits
Achievement
Complete all 6 lessons to finish the Sequence Models (RNNs & LSTMs) module.