Compare every algorithm
One click ranks all algorithms on the same split. This is a real run on iris.
| Algorithm | Accuracy |
|---|---|
| Support Vector Machine | 0.9667 |
| Logistic Regression | 0.9333 |
| Decision Tree | 0.9333 |
| K-Nearest Neighbors | 0.9333 |
| Random Forest | 0.9000 |
Watch a model learn, one stage at a time.
A hands-on scikit-learn pipeline where every step is open: split, preprocess, train, evaluate. Break it, compare it, understand it.
k-nearest neighbours, running in your browser
accuracy 100.0%
Click the plot to add points. Red rings mark points the model would get wrong.
Six stages, all inspectable.
Run one stage at a time and read what it produced, or run the whole pipeline in a click.
Choose a built-in dataset or upload a CSV. Column types are detected as soon as it loads.
- dataset:
- iris
- rows:
- 150
- problem_type:
- classification
Everything a real workflow needs.
Decision boundaries
Pick any two features and see where the trained model draws its lines.
5-fold
Grid search tuning
Cross-validated, with the best parameters and score shown.
Bring your own CSV
Numeric and categorical columns are detected and handled for you.
Ask the assistant
What is overfitting?
Overfitting happens when a model learns the training data too well, including its noise and quirks, rather than the underlying pattern.
Correct where it counts.
No data leakage
The split happens first. Preprocessing is fitted on training rows and only applied to the test set.
One path for every dataset
Built-in or uploaded, each dataset gets a schema that drives preprocessing and algorithm choice.
A fair comparison
Every algorithm trains on the same split and is scored on the same test rows.
backend/main.py
X_train, X_test, y_train, y_test = train_test_split(
X, y, test_size=0.2, random_state=42, stratify=y
)
preprocessor = build_preprocessor(X_train)
X_train_t = preprocessor.fit_transform(X_train)
X_test_t = preprocessor.transform(X_test)The test set is only ever transformed, never fitted on.
Seven algorithms. Five datasets. Or yours.
- Linear Regression
- Logistic Regression
- Decision Tree
- Random Forest
- K-Nearest Neighbors
- Support Vector Machine
- Neural Network
- irisclassification
- wineclassification
- breast_cancerclassification
- diabetesregression
- california_housingregression
Run your first pipeline in under a minute.
No sign-up needed. The backend sleeps when idle, so the first load can take a few seconds.