Hotelling
Dion Research's Hotelling package now has control charts for quality control, anomaly detection and multivariate EDA. This is in addition to the T2 (T-squared) statistics.
The v0.3 of the package is available on pypi: https://pypi.org/project/hotelling/
As far as we know, this is the only implementation that covers both UCL and LCL. Without the lower control line, in the above example, the fifth multivariate sample (index 4) would not have been identified as out of control (anomalous / outlier / exotic).
A straightforward example is available in a jupyter notebook on github.
Francois Dion
Chief Data Scientist
@f_dion
About Dion Research LLC: We are a boutique Data Science consultancy . established 2011. As we do end-to-end Data Science, we can help you solve business problems every step of the way. Get in touch for more information.
The article introduces version 0.3 of Dion Research's Hotelling package, which adds multivariate control charts for quality control, anomaly detection, and multivariate exploratory data analysis. The package also includes T2 (T-squared) statistics, providing additional capabilities for analyzing multivariate observations and identifying unusual samples.
ReplyDeleteThe article mentions that a straightforward example of the multivariate control chart functionality is available through a Jupyter Notebook on GitHub. Such examples can help users understand how statistical techniques can be applied to real datasets and how analytical results can be examined visually. These concepts are relevant to Data Analysis Course.
ReplyDeleteMultivariate control charts can provide a useful way to examine patterns, detect anomalies, and support quality-control analysis when several variables need to be considered together. Presenting these analytical results clearly is also an important part of the workflow, making Data Visualization Course relevant when developing visual analytical reports and dashboards.
ReplyDeleteA key feature highlighted in the article is support for both upper and lower control limits (UCL and LCL). Having both control lines can be important when identifying observations that fall outside the expected range, as demonstrated by the example where the fifth multivariate sample is identified as an anomalous or out-of-control observation.Data Science Projects for Final Year
ReplyDelete