Data Sources

MathExec supports multiple ways to get data onto the canvas. All data is loaded as tabular (CSV-like) data for training.

CSV Upload

The most common method. Click the Data tool (D) on the canvas or use the dataset panel to upload a CSV file from your computer.

• Max 10,000 rows
• First row is treated as column headers
• Numeric and categorical columns are auto-detected

Sample Datasets

MathExec includes built-in sample datasets for quick testing:

• Binary Classification — 2 features, binary target (great for logistic regression)
• More samples are added regularly

Kaggle

Import datasets directly from Kaggle. Enter a Kaggle dataset URL or search by name. MathExec downloads and converts the data to tabular format.

HuggingFace

Import datasets from the HuggingFace Datasets hub. Paste a HuggingFace dataset URL and MathExec fetches the data.

UCI Machine Learning Repository

The classic ML benchmark repository. Import any UCI dataset by URL.

URL Import

Paste a direct link to any publicly hosted CSV file. MathExec downloads and parses it.

• Must be a direct link to a .csv file (not an HTML page)
• CORS must be enabled on the host, or the file will be fetched via the backend proxy

Data Management

All imported datasets appear in the Dataset panel. From there you can:

• Select — activate a dataset for use on the canvas
• Delete — remove a dataset from your workspace
• Preview — open the data table with sorting and column stats

Datasets are persisted in your browser's localStorage. They survive page refreshes but are specific to your browser.