CBM+ Text Mining & Analytics
A vendor requested to showcase data science and text mining/analytics capabilities in a real-time demo in support of desired Condition Based Maintenance (CBM+).
Illumination Works applied modern text mining, advanced machine learning, and sound statistics to quickly analyze, mine, cleanse, merge, explore, and expose maintenance text data for CBM+ insights.
- Improved overall accuracy of enumerated codes from 70% to 90%
- Applied advanced text analytics to look at correlations between free text and categorical variables
- Short proof of concept showed vast value in text mining methodology
- Provided valuable feedback to the maintainer on the quality and descriptiveness of maintenance notes
- Python, R, Shiny
- Bag of Words (BoW) model to establish word relationships
- Machine learning to deduce meaning from free text
- Statistical inference to predict likelihood of occurrence