One of the beauties of Mapper is that it can be used on wide variety of data and problems. Here we highlight a few cases.
KeplerMapper can be used for anomaly detection. Anomaly detection is useful for fighting fraud and finding errors. Projecting with
knn_distance_5 can surface outliers. There are also specialized outlier detection algorithms, like the Isolation Forest and GLOSH, that make good projections for anomaly detection.
Notebooks: TDA Wisconsin Breast Cancer
Demos: TDA Wisconsin Breast Cancer
Natural Language Processing¶
Mapper can be used to explore bias in Word2Vec models.
Notebooks: Word Vector Gender Bias.
Demos: Word Vector Gender Bias
Uses & Mentions¶
Svetlana Lockwood presented Kepler-Mapper at an ACM Workshop Open Source Software for TDA
Abbas H Rizvi et al. mention Kepler-Mapper in the Nature paper Single-cell topological RNA-seq analysis reveals insights into cellular differentiation and development
Jin-Ku Lee et al. mention Kepler-Mapper in the Nature paper Spatiotemporal genomic architecture informs precision oncology in glioblastoma
Natalino Busa maps credit risk Predicting Defaulting on Credit Cards
Mark Coudriau et al. maps the Darknet Topological analysis and visualisation of network monitoring data: Darknet case study
Steve Oudot et al. from École Polytechnique give a course on TDA using Kepler-Mapper: INF556 — Topological Data Analysis
George Lake et al. from Universität Zürich use it for teaching data science: ESC403 Introduction to Data Science
Mikael Vejdemo-Johansson maps process control systems TDA as a diagnostics tool for process control systems.
Jeffrey Ray et al. review KeplerMapper in Advances in Intelligent Networking and Collaborative Systems: A Survey of Topological Data Analysis (TDA) Methods Implemented in Python
Wei Guo et al. succesfully use KeplerMapper in prediction of manufacturing systems: Identification of key features using topological data analysis for accurate prediction of manufacturing system outputs
Christian Parsons uses KeplerMapper in Mapamundi: a world map based on a geometric interpretation of data, it takes socio-economic metrics for all countries and uses T-SNE as lens and DBSCAN as clusterer to make an alternative world map. Awarded on the World Data Visualization Prize 2019