Producer risk in manufacturing is the probability that a good product will be rejected by a consumer because a bad batch was observed in a sample that has been tested. Due to the cost of manual quality checks, often times we rely on checking the quality of products in samples, and consecutive samples and make a decision whether a lot of produced/manufactured items are acceptable to market/consumption.
Here we replicate the contributions in chain sampling plans from research in academia and make them available for consumption.
In this project, we delve deep into the chain sampling method, a pivotal technique in quality control. Using synthetic data, we simulate two distinct real-world scenarios:
- Manufacturing of sneaker insoles
- Production of LED bulbs
Our objective is to decode the criteria for batch acceptance or rejection based on defect or failure rates.
-
Chain Sampling Analysis Notebook: This Jupyter notebook contains a detailed analysis, from generating synthetic data to visualizing results. It's equipped with explanations at each step to ensure clarity of the approach and findings.
-
Sampling Code: The core sampling functions and methodologies are derived from the code provided in the
sampling_min_sum.pyfile.
-
Synthetic Data Generation: We create realistic synthetic datasets for both scenarios, providing a foundation for our analysis.
-
Sampling Plans: Utilizing the single sampling plan function, we determine the acceptance or rejection of batches.
-
Visualization: A visual representation of defect rates and decisions, aiding in a clearer understanding of the results.
- Clone the repository to your local machine.
- Navigate to the directory and open the
chain_sampling_analysis.ipynbnotebook. - Execute the notebook cells sequentially to understand the analysis flow.
- Python 3.8+
- Libraries: pandas, numpy, matplotlib
Feel free to fork this project and enhance it. Pull requests with improvements and optimizations are always welcome.