process-improvement 0.1.0

A package for process capability analysis

process_improvement.py

The process_improvement.py library is a collection of functions designed to help identify, undersand, and eliminate the influence of the two types of variation from business and manufacturing processes. While the primary tool by which this is facilitated is the the process behavior chart (otherwise known as a control chart), process_improvement.py contains additional modules and functions related to the task of process improvement. The process_improvement.py library is part of the larger body of work called The Broken Quality Initiative (BrokenQuality.com). The aim of BQI is to address industries' pervasive lack of knowledge regarding variation and the only tool capable of making sense of variation, the process behavior chart (control chart). .

Visit BrokenQuality.com for resources and more details regarding the application and use of process behavior charts.

Table of Contents

• Prerequisites
• Installation
• Usage
• Modules
• Contributing
• Contact
• License
• Additional Information

Prerequisites

Before you begin, ensure you have met the following requirements:

• You have installed Python 3.6 or higher.
• You have a working knowledge of Python and data analysis libraries such as pandas and matplotlib.
• You have a working knowledge of Process Behavior Charts and Statistical Process Control.

Installation

To install process_improvement.py directly from GitHub, enter the following command using the command prompt:

pip install git+https://github.com/jimlehner/process_improvement

Usage

After installation the process_improvement.py library can be used as follows:

1. From process_improvement import xmr_charts module as xmr: from process_improvement import xmr_charts as xmr
2. Call the function of interest from the requisite module: xmr.xmrchart(df, 'Values', 'Observations', title='Example X-chart')

Modules

The process_improvement.py pacakge contains 5 modules:

1. xmr_charts
2. process_capability
3. comparison_charts
4. limit_charts
5. network_analysis Each of these modules can be used to address a different aspect of understanding variation.

Functions

xmr_charts.py The xmr_charts module contains 3 function:

1. xmrchart: Generates a process behavior chart of individual values and a moving range called an XmR Chart from the provided DataFrame.
2. xchart: Generates the X Chart portion of an XmR Chart from the provided DataFrame.
3. mrchart: Generates the moving range (mR) Chart portion of an XmR Chart from the provided DataFrame.

process_capability The process_capability module contains 3 functions:

1. capability_histogram: Generates a capability histogram of the provided process data in the context of the specifciation limits and the option of displaying the process capability indices.
2. multi_chart: Generates the X Chart portion of an XmR Chart and a capability histogram in the same figure to enable direct visual comparison of process behavior and the distribution of the data.
3. process_capabilities: Calculates the process capability indices of Cp, Cpk, Pp, and Ppk, based on the provided process data, upper and lower specification limits, and target value.

comparison_charts The comparison_charts module contains 3 functions:

1. xchart_comparison: Generates X Charts from the provided list of DataFrames and visually compares their statistics.
2. mrchart_comparison: Generates moving range (mR) Charts from the provided list of DataFrames and visually compares their statistics.
3. xmr_comparison: Generates a 2x2 subplots of XmR Charts from the provided list of DataFrames and visually compares their statistics.

limit_charts The limit_chart module contains 1 function:

1. limit_chart: Generates a limit chart that sequentially plots process data in the context of the provided specification limits and target.

network_analysis The network_analysis module contains 2 function:

1. network_analysis: Generates a list of small multiples each containing the X Chart portion of an XmR Chart from the provided list of DataFrames. For an example of how to use network_analysis see the essay Network Analysis: Advancing the utility of SPC
2. limit_chart_network_analysis: Generates a grid of small multiples composed of the list of dataframes provided by df_list.

Notes

If you are unfamiliar with process behavior charts (control charts) visit BrokenQuality.com.

Contributing

To contribute to DataDrivenImprovement, follow these steps:

1. Fork this repository.
2. Create a branch: git checkout -b <branch_name>.
3. Make your changes and commit them: git commit -m '<commit_message>'
4. Push to the original branch: git push origin <DataDrivenImprovement>/<location>.
5. Create the pull request.

Alternatively see the GitHub documentation on creating a pull request.

Contact

If you want to contact me you can reach me at [e-mail].

License

This project uses the following license: MIT License.

Additional Information

• Parts of a Process Behavior Chart: Invented by Dr. Walter Shewhart in the mid-1920s at Bell Laboratories, PBCs are composed of two charts: the X-chart and the mR-chart. Where the X-chart bounds the variation associated with individual values the mR-chart bounds the value-to-value variation. This is made possible through the calculation of a trio of limits known as process limits. The upper process limit (UPL) and lower process limit (LPL) are used on the X-chart. The upper range limit (URL) is used on the mR-chart.
• Two types of variation: Inherent in the characterizations of predictable and unpredictable is the tyoe of variation action a process. A predictable process is influenced by only routine causes of variation. An unpredictable process is influenced by both routine causes of variation and assignable causes of variation.
• Improvement:
  • Predictable: To improve a predictable process routine causes of variation must be identified, understood, and mitigated. This requires fundamental changes to the process must be made. These include, but are not limited to, changes to raw materials, adjustment to system settings, redesign of stations, redesign of software, calibration of measurement systems.
  • Unpredictable: To improve an unpredictable process assignable causes of variation must be identifed, understood, and eliminated. To begin this process, an investigation into values that fall outside the process limits on the PBC must be performed.
• For those unfamiliar with process behavior charts (control charts) that are interested in learning more visit BrokenQuality.com.