regressionmadesimple 4.1.0.dev0

Minimalist machine learning toolkit that wraps `scikit-learn` for quick prototyping. Just `import rms` and go.

RegressionMadeSimple v4.1.0 🚀

A minimalist ML toolkit wrapping scikit-learn for quick prototyping. Just import regressionmadesimple as rms and go!

---

What's New in v4.1.0 🎉

🧪 Experiment Workflow (New)

A stateful rms.Experiment class that manages the full ML pipeline — smart column typing, multi-split management, scaler integration, model deduplication, and full state persistence via dill.

import regressionmadesimple as rms
import pandas as pd
from sklearn.linear_model import LinearRegression

df = pd.DataFrame({
    "x": [1, 2, 3, 4, 5],
    "cat": ["a", "b", "a", "b", "a"],
    "y": [2.1, 4.2, 6.1, 8.3, 10.2]
})

exp = rms.Experiment(df, target="y", out_path="./experiment")
exp.fit_models({
    "lr": [LinearRegression()],
})

# Explore
results = exp.model_results["lr_0"]
print(results["metrics"]["test"]["r2_score"])

# Persist everything
exp.save()
loaded = rms.Experiment.load("./experiment/experiment.dill")

Key features:

• Smart column typing — bool, 0/1-as-bool, numeric, categorical auto-detected
• Multi-split management — multiple train/test/validation splits with metadata
• Scaler integration — MinMaxScaler (default) or StandardScaler
• Validation split — 70/20/10 three-way splits
• Model dedup — identical configs auto-skipped
• Rich metrics — r², MSE, RMSE, MAE, MAPE, explained variance + classification metrics
• dill persistence — save/load full experiment (data + models + metadata)
• add_columns() — propagate new features to all existing splits retroactively
• Classifier support — works with LogisticRegression, KNN, etc.

🧹 Cleanup

• Removed legacy curves.py and base_class.py (dead code since v3)
• Fixed scaler dtype warnings

---

Installation

pip install regressionmadesimple

Or from source:

git clone https://github.com/Unknownuserfrommars/regressionmadesimple.git
cd regressionmadesimple
pip install -e .

---

Quick Start

Experiment API (v4.1.0)

import regressionmadesimple as rms
import pandas as pd
from sklearn.linear_model import LinearRegression
from sklearn.ensemble import RandomForestRegressor

# Load data
data = pd.DataFrame({
    "x1": range(50),
    "x2": __import__("numpy").random.randn(50),
    "cat": __import__("numpy").random.choice(["a", "b", "c"], 50),
    "y": [2*v + __import__("numpy").random.randn()*0.5 for v in range(50)],
})

# Create experiment
exp = rms.Experiment(data, target="y", out_path="./my_exp")

# Fit multiple models at once
exp.fit_models({
    "lr": [LinearRegression()],
    "rf": [RandomForestRegressor(n_estimators=50, random_state=42)],
})

# Compare results
for key, res in exp.model_results.items():
    r2 = res["metrics"]["test"]["r2_score"][0]
    print(f"{key}: R² = {r2:.4f}")

# Save everything
exp.save()

Classic Model API (v4.x)

import regressionmadesimple as rms

data = pd.DataFrame({
    "x": [1, 2, 3, 4, 5],
    "y": [2.1, 4.2, 6.1, 8.3, 10.2]
})

model = rms.models.Linear(data, "x", "y")
print(f"R²: {model.r2_score():.4f}")
print(f"RMSE: {model.rmse():.4f}")

model.save_model("my_model.pkl")

---

Available Models

Model	Class	Description
Linear	rms.models.Linear	Simple linear regression (y = mx + b)
Quadratic	rms.models.Quadratic	Polynomial regression (degree=2)
Cubic	rms.models.Cubic	Polynomial regression (degree=3)
Custom Curve	rms.models.CustomCurve	Custom basis functions

---

Roadmap 🗺️

See ROADMAP.md for the full development plan covering v4.x through v7.x.

---

Contributing

PRs welcome!

1. Fork the repo
2. Create your feature branch (git checkout -b feature/AmazingFeature)
3. Commit your changes (git commit -m "Add some AmazingFeature")
4. Push to the branch (git push origin feature/AmazingFeature)
5. Open a Pull Request

---

License

MIT — see LICENSE.

---

Changelog

v4.1.0-dev (2026-05-15)

• ✨ New: rms.Experiment — full experiment workflow with smart column typing, multi-split management, scaler integration, model dedup, dill persistence
• ✨ New: _LOWER_IS_BETTER — exported frozenset for smart metric sorting
• 🧹 Removed: Legacy curves.py and base_class.py (dead code)

v4.0.0 (2025-01-01)

• Model registry pattern (rms.models.*)
• Enhanced BaseModel with common functionality
• Model serialization (save_model(), load_model())
• Additional scoring metrics (r2_score(), mae(), rmse())
• String-based model specification removed

v3.0.0

• Refactored codebase, model registry, deprecation warnings

v2.0.0

• Initial public release

---

Made with ❤️ by Unknownuserfrommars