gri-fitter 0.2.2

Structured wrapper around scipy.optimize.curve_fit with automatic coefficient extraction

Fitter (Equation Fitting)

Structured wrapper around scipy.optimize.curve_fit that pairs equations with their fitted results. Provides automatic coefficient extraction from function signatures, printable equation representations with value substitution, and RMSE reporting.

Overview

gri-fitter separates equation definition from curve fitting into two composable classes. The Equation class wraps any Python function and introspects its signature to identify the independent variable (first parameter) and coefficients (remaining parameters). The Fit class takes data and an Equation, runs scipy.optimize.curve_fit, and stores the results -- coefficients, fitted y-values, and RMSE -- for downstream use.

Requires Python 3.12+.

Installation

pip install gri-fitter

For development:

git clone https://gitlab.com/geosol-foss/python/gri-fitter.git
cd gri-fitter
. .init_venv.sh

Quick Start

from gri_fitter import Equation, Fit

# Define the equation: first param is independent variable, rest are coefficients
equ = Equation(lambda x, m, b: m * x + b, "mx + b")

# Fit to data
fit = Fit((0, 1, 2, 3), (0, 1.1, 1.9, 3), equ, "LineFit")
coeffs, rmse = fit.fit()

# Print results
print(fit)          # LineFit RMSE [0.00900] ::: 0.99x + 0.05
print(fit.y_fit)    # Fitted y values at the input x points

Equation Class

Equation wraps a function (lambda or def) and extracts variable names from its signature. The first parameter is treated as the independent variable; all others are coefficients.

# Using a lambda
equ = Equation(lambda x, a, b, c: a * x**2 + b * x + c, "ax^2 + bx + c")

print(equ.independent)   # "x"
print(equ.coefficients)  # ["a", "b", "c"]

# Using a defined function
def exponential(x, a, k):
    return a * np.exp(k * x)

equ = Equation(exponential, "a * exp(kx)")

# Without a string representation (replace() will not be available)
equ = Equation(lambda x, m: m * x)

The replace() method substitutes coefficient values into the equation string for display. The p parameter controls decimal precision (truncation, not rounding):

equ = Equation(lambda x, m, b: m * x + b, "mx + b")
print(equ.replace((4.9876, 1.0)))         # 4.987x + 1.0
print(equ.replace((4.9876, 1.0), p=1))    # 4.9x + 1.0

Fit Class

Fit takes x data, y data, an Equation, and an optional title. Call fit() to run scipy.optimize.curve_fit.

equ = Equation(lambda x, m, b: m * x + b, "mx + b")
fit = Fit(x_data, y_data, equ, "MyFit")

# Returns (coefficients, rmse) on success, ([], None) on failure
coeffs, rmse = fit.fit()

# After fitting, results are stored as attributes
print(fit.coefficients)  # [0.99, 0.05]
print(fit.rmse)          # 0.009
print(fit.y_fit)         # Fitted y values at each input x

When the fit fails (e.g., data does not match the equation form), fit() returns ([], None) and str(fit) displays the failure:

print(fit)  # MyFit ***NO FIT*** ::: mx + b

Dependencies

• numpy: Array operations
• scipy: scipy.optimize.curve_fit for nonlinear least squares fitting

Other Projects

Current list of other GRI FOSS Projects we are building and maintaining.

License

MIT License. See LICENSE for details.