https://github.com/nishatrhythm/machine-learning
A dynamic repository showcasing practical Machine Learning projects, featuring cutting-edge techniques, model training, and hyperparameter optimization for impactful insights.
https://github.com/nishatrhythm/machine-learning
hyperparameter-optimization machine-learning
Last synced: over 1 year ago
JSON representation
A dynamic repository showcasing practical Machine Learning projects, featuring cutting-edge techniques, model training, and hyperparameter optimization for impactful insights.
- Host: GitHub
- URL: https://github.com/nishatrhythm/machine-learning
- Owner: nishatrhythm
- License: mit
- Created: 2025-01-13T06:38:42.000Z (over 1 year ago)
- Default Branch: main
- Last Pushed: 2025-01-16T06:24:11.000Z (over 1 year ago)
- Last Synced: 2025-01-16T07:31:30.841Z (over 1 year ago)
- Topics: hyperparameter-optimization, machine-learning
- Language: Jupyter Notebook
- Homepage:
- Size: 1.93 MB
- Stars: 0
- Watchers: 1
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
# Hyperparameter Optimization for LSTM on Daily Minimum Temperatures Dataset
This repository demonstrates hyperparameter optimization for Long Short-Term Memory (LSTM) networks on the **Daily Minimum Temperatures Dataset**. The objective is to predict the next day's temperature using time-series data and evaluate the impact of various hyperparameters on the model's performance.
## Dataset
The dataset used is the **Daily Minimum Temperatures Dataset**, which contains daily minimum temperatures recorded in Melbourne, Australia. It spans from 1981 to 1990 and can be directly downloaded from [this link](https://raw.githubusercontent.com/jbrownlee/Datasets/master/daily-min-temperatures.csv).
## Features
- **Dynamic hyperparameter optimization** using combinations of:
- Activation Functions: `tanh`, `relu`
- Number of Neurons: `50`, `100`
- Dropout Rates: `0.2`, `0.3`
- Number of Layers: `1`, `2`
- Optimizers: `adam`, `rmsprop`
- Learning Rates: `0.001`, `0.01`
- Epochs: `20`, `30`
- Results logged into CSV files:
- `hyperparameter_results.csv`: Contains metrics for all hyperparameter combinations.
- `best_hyperparameters.csv`: Contains the best-performing hyperparameter configuration.
- Visualization of training and validation losses for the best model.
- **Best model saved** as `best_lstm_model.h5` for future use.
## Execution Environment
This project was executed on **Google Colab** using the **v2-8 TPU**. The total execution time was approximately **5378.465 seconds**.
## Files
- **`hyperparameter_results.csv`**: Contains performance metrics (MSE, MAE) for all hyperparameter combinations.
- **`best_hyperparameters.csv`**: Contains the best-performing hyperparameter configuration based on MSE.
- **`best_lstm_model.h5`**: The trained LSTM model using the best hyperparameters.
## Results
The best-performing hyperparameters are as follows:
| Activation Function | Number of Neurons | Dropout Rate | Number of Layers | Optimizer | Learning Rate | Epochs | Test Loss | MAE | MSE |
|---------------------|-------------------|--------------|------------------|-----------|---------------|--------|-----------|--------|--------|
| `tanh` | 100 | 0.2 | 2 | `rmsprop` | 0.01 | 20 | 0.007019 | 0.0659 | 0.0070 |
## Performance
- **Test Loss**: `0.007019`
- **Mean Absolute Error (MAE)**: `0.0659`
- **Mean Squared Error (MSE)**: `0.0070`
## Visualization
The graph shows the training and validation loss for the best model over epochs.
## References
- Dataset: [Daily Minimum Temperatures Dataset](https://github.com/jbrownlee/Datasets)
- LSTM implementation and hyperparameter optimization techniques.