spotdesirability implements the Derringer & Suich (1980) desirability framework, providing a flexible set of tools for multi-objective optimization through desirability functions. spotdesirability is a Python implementation of the R package desirability, which provides S3 classes for multivariate optimization using the desirability function by Derringer and Suich (1980). The R desirability package is maintained and developed by Max Kuhn. It is is available on CRAN: https://CRAN.R-project.org/package=desirability, DOI: https://doi.org/10.32614/CRAN.package.desirability
The conversion_pred function implements a second-order Response Surface Methodology (RSM) model. It predicts percent conversion based on three coded factors: x1 (Reaction Time), x2 (Temperature), and x3 (Amount of Catalyst).
The model equation is: y = 81.09 + 1.03x1 + 4.04x2 + 6.20x3 - 1.84x1^2 + 2.94x2^2 - 5.19x3^2 + 2.22x1x2 + 11.38x1x3 - 3.88x2x3
The following living code block demonstrates how to use the function with numpy arrays:
import numpy as np
from spotdesirability.functions.rsm import conversion_pred
# Define input parameters [x1, x2, x3]
x = np.array([1.0, 0.5, -0.2])
# Calculate predicted conversion
pred = conversion_pred(x)
print(f"Predicted conversion for {x}: {pred:.2f}%")Predicted conversion for [ 1. 0.5 -0.2]: 80.81%