uncertainty-tools

Utilities for uncertainty-aware regression and symbolic function helpers.

Python version: 3.10+

📦 PyPI package name: uncertainty-tools
🧩 Python import name: unc_tools

---

Installation

Using pip

pip install uncertainty-tools

Using uv

uv pip install uncertainty-tools

---

Notebooks

link

---

Usage

from unc_tools import FunctionBase1D, Poly, UncRegression

# Symbolic expression with coefficients
expr = FunctionBase1D("a*x + b")

# Polynomial helper
poly = Poly(2)

# Regression with uncertainty propagation
reg = UncRegression(
    x=[0, 1, 2],
    y=[0, 1.1, 1.9],
    func=expr,
)

pred = reg.predict([0.5, 1.5])

---

Optional matplotlib and sympy patches

Importing unc_tools.patches monkey-patches:

• matplotlib.axes.Axes.scatter

• matplotlib.axes.Axes.plot

and adds uncertainty-aware helpers for SymPy substitution and lambdify.

⚠️ This introduces global side effects and must be enabled explicitly:

import unc_tools.patches  # noqa: F401

---

Public API

• UncRegression — regression with uncertainty propagation
• FunctionBase1D, Poly, Hyper — symbolic helpers for 1D expressions
• DataError, ExpressionError, InitialGuessError, ModelTypeError — custom exceptions

---

Project status

⚠️ This project is under active development.
The public API may change between minor versions.

---

License

This project is licensed under the MIT License.