Learn machine learning by moving through a clear, honest workflow.
Train, compare, explain, export, and deploy scikit-learn models in three lines.
Train your first model in minutes.
| 1 | from breezeml import Breeze |
| 2 | |
| 3 | data = Breeze.load("data.csv") |
| 4 | |
| 5 | model = Breeze.setup( |
| 6 | model="xgboost" |
| 7 | ) |
| 8 | model.train(data) |
| 9 | model.evaluate() |
Clean, encode, scale - visualize the flow.
Split 80/20See what works best - honestly.
| Model | ROC AUC | Accuracy | F1 | Time |
|---|---|---|---|---|
 XGBoost | 0.912 | 0.876 | 0.875 | 0.74s |
| LightGBM | 0.907 | 0.872 | 0.871 | 0.66s |
| Random Forest | 0.893 | 0.859 | 0.858 | 0.58s |
| Logistic Regression | 0.842 | 0.803 | 0.801 | 0.12s |
| SVM (RBF) | 0.831 | 0.792 | 0.790 | 1.21s |
Ship your model with confidence.
Why not just scikit-learn?
BreezeML chooses the task, preprocessing, split, metrics, and cross-validation defaults so the first model is honest.
compare(), explain_decisions=True, audit(), fairness, drift, and cards make each result easier to inspect.
export() writes a standalone scikit-learn script, so the trained workflow can leave BreezeML behind.
