Stock Price Prediction using Sequence Model
Forecasting using LSTM
Learning Outcome
5
Reverse the scaling to interpret predictions in real dollar values
4
Train (fit) the model on your prepared 3D data
3
Compile the model with the correct optimizer and loss function
2
Add an LSTM layer configured for time series data
1
Build a Sequential neural network architecture in Keras
Recall: The Missing Piece
Theory & Scaling
Scaled data between 0 and 1 (MinMax)
60-Day Slicing
Lookback windows created
3D Reshape
Tensor shape: (Samples, 60, 1)
Fueling
THE ENGINE
Keras LSTM Model
Ready to Build
Transition from Analogy to Technical Concept(Slide 5)
Core Concepts (Slide 6)
Core Concepts (Slide 7)
Core Concepts (.....Slide N-3)
Summary
5
6
.fit() the model on the training data.
.fit() the model on the training data.
.predict() the test data and apply inverse_transform().
4
.compile() with Adam optimizer and MSE loss.
3
Add Dense(1) layer for the final output.
2
Add LSTM() layer with correct input_shape.
1
Initialize Sequential() model.
Quiz
Why do we add a Dense(units=1) layer at the very end of our LSTM network for stock prediction?
A) To increase the memory of the LSTM
B) To output a single continuous value (the predicted price)
C) To convert the data into a 3D tensor
D) To scale the data between 0 and 1
Quiz-Answer
Why do we add a Dense(units=1) layer at the very end of our LSTM network for stock prediction?
A) To increase the memory of the LSTM
B) To output a single continuous value (the predicted price)
C) To convert the data into a 3D tensor
D) To scale the data between 0 and 1
