CS456MLModelPlatform

A browser-based tool for uploading datasets, training machine learning models, and viewing evaluation metrics — all client-side with no installation required.

Live Repository: https://github.com/YourUsername/Web-ML-Platform

Using the ML Model Platform

Environment Requirements

This project is designed to run entirely client-side in a modern web browser. No installation, internet connection, or external dependencies are required.

• Operating System: Windows, macOS, or Linux (any recent version)
• Browser: Chrome, Edge, Firefox, or Safari (latest version recommended)
• Storage: The platform uses the browser’s localStorage feature to save user accounts and training history. Clearing browser data will erase saved models.

Running the Project

Running from Zip file

1. Download the WebMLPlatformDeliverables zip folder from Blackboard or clone from Github repository
2. Open the folder and locate index.html within the WebML folder
3. Run the file in the browser of your choice

Running with Docker

1. Download the WebMLPlatformDeliverables zip folder from Blackboard or clone from Github repository
2. cd to WebMLPlatformDeliverables and run docker compose up
3. Open your browser to http://localhost:8080

Creating an Account

1. Click Create account if you are new (If returning user, skip to step 7)
2. Enter your new username, and your new password, then confirm password
3. Click create account!
4. If not auto-directed back to login page, click "back to login"
5. Sign in using your username and password, then click "Login"

Training a Dataset

1. Navigate to "Upload Dataset" and click "Click to upload"
2. Select the data you wish to process! (There are two sample data sets in the zip file called classification_dataset_small.csv and linear_regression_data.csv respectively)
3. Use the drop down menu located under "Select Target Variable" and select your target
4. Under "Select Models", click the models you would like to use
5. Click train models, the site will train the data on all of the models that are highlighted.
6. After loading, you will be able to see your results on the right side of the screen

Utilizing "History"

1. Scroll to the bottom of the results page, and if you wish to save the model for later, click "Save to History"
2. To see history, navigate to "History" using the button located in the top right
3. From here, you can use "View Results" to see the model on the "Train Models" screen, or click "Delete" to remove the model from history

Logging out

1. Now, if you wish to log out, you can navigate to the "Logout" button in the rightmost upper corner.

Reproducing Our Reported Results

To reproduce our results exactly:

1. Upload classification_dataset_small.csv
2. Select Label as the target variable
3. Train using Logistic Regression and SVM
4. You should see results similar to:
   • Logistic Regression:
     • Accuracy ≈ 0.90
     • AUC ≈ 0.9625
   • SVM:
     • Accuracy ≈ 0.90
     • AUC ≈ 0.9631 (Minor variations due to randomness are expected.)

Responsible AI & Data Handling

• This app processes all data locally in your browser — nothing is uploaded or stored on external servers.
• Sample datasets are AI generated and contain no personally identifiable information (PII).
• Users are responsible for ensuring that uploaded data complies with privacy and ethical standards.
• Results are for educational and experimental use only, not for real-world decision-making.