This repository contains the Python code to reproduce the results of the paper "Effective Non-Random Extreme Learning Machine". Follow the instructions below to set up and run the code.
To ensure compatibility and proper functionality, use Python 3.10. Install the required Python packages as listed below:
numpy
scipy
matplotlib
scikit-learn
Install the packages using the following command:
pip install numpy scipy matplotlib scikit-learn
Before running the code, download the real datasets and place them in a folder named datasets
. If this folder does not exist, create it manually.
Download the datasets from the following links and organize them accordingly:
- Abalone - Download Link
- Auto MPG - Download Link
- California Housing - Download Link
- Delta Ailerons - Download Link
- LA Ozone - Download Link (Save as
datasets/LAozone/LAozone.data.txt
) - Machine CPU - Download Link
- Prostate Cancer - Download Link (Save as
datasets/prostate/prostate.data.txt
) - Servo - Download Link
Before running any experiments, it is recommended to execute datagenerator.py
. This script will generate the synthetic datasets used in the experiments.
Run the following command:
python datagenerator.py
The repository includes two Jupyter notebooks:
experiments.ipynb
– This notebook runs the experiments and produces.csv
files containing the results.presentation.ipynb
– This notebook reads the results fromexperiments.ipynb
and generates tables and plots for analysis.
To create a plot with zoom that displays training and test errors for all models on real datasets, run dynamicplot.py
:
python dynamicplot.py
This interactive plot allows you to zoom in on a specified x-axis range, which is useful for detailed analysis and was used to create Figure 4 of the paper. If zooming is not required, you can produce the plot directly using presentation.ipynb
.