Economic Design and Optimization of CSP-1 Continuous Sampling Plans

Reproducibility package for the paper:

Economic Design and Optimization of CSP-1 Continuous Sampling Plans with a Four-Component Cost Model under Markovian Transition Dynamics Author names omitted for double-anonymized peer review. Submitted to the International Journal of Production Economics.

This repository contains the model, optimizer, simulation, sensitivity/robustness analyses, and the empirical validation on a real semiconductor stream (SECOM) used in the paper. The manuscript itself is not included.

Contents

File	Description
csp_model.py	Four-component CSP-1 economic cost model (CSP1EconomicModel) and optimize_csp1, derived by renewal-reward analysis of a two-state Markov chain.
optimize_economic.py	Driver that optimizes the clearance number i (exhaustive search) and sampling fraction f (Brent's method).
simulate_csp1.py	Discrete-event simulation of the CSP-1 rule, used to validate the analytic model (agreement within 1%).
validate_optimizer.py	Cross-checks the optimizer against the simulation.
verify_consistency.py	Consistency checks (e.g. AFI vs. the classical CSP-1 characteristic).
empirical_secom.py	Estimates p' from the real SECOM stream and validates each policy by replay over the true defect sequence.
run_sensitivity.py, run_sensitivity_extended.py	Sensitivity of the optimum to the cost parameters and p'.
run_robustness.py	Robustness of incurred cost to misestimation of p'.
generate_results_table.py, generate_heatmaps.py	Reproduce the results tables and heatmaps.
data/secom_labels.data	Pass/fail labels of the SECOM data set, in temporal order (see Data sources).
results/*.csv	Pre-computed numerical outputs (regenerable by the scripts above).

Requirements

pip install -r requirements.txt

Python 3.10+ with numpy, scipy, pandas, matplotlib.

Reproducing the results

Run from the repository root (scripts resolve paths relative to their own location):

python optimize_economic.py        # economic optimum
python simulate_csp1.py            # discrete-event validation
python verify_consistency.py       # AFI / model consistency checks
python run_sensitivity_extended.py # sensitivity study
python run_robustness.py           # robustness to p' misestimation
python empirical_secom.py          # SECOM estimation + replay validation
python generate_results_table.py   # results tables
python generate_heatmaps.py        # heatmaps

Data sources

The SECOM data set is publicly available from the UCI Machine Learning Repository:

McCann, M., Johnston, A. (2008). SECOM Data Set. UCI Machine Learning Repository. https://doi.org/10.24432/C54305

Only the pass/fail label file (secom_labels.data), in its original temporal order, is redistributed here for convenience; it is used to estimate the empirical defect rate and to replay the inspection policies over the real stream.

License

The code in this repository is released under the MIT License (see LICENSE). The SECOM data set retains its original UCI license and citation.