What if the decision-making process was heavily reliant on existing data? What if this data does not exist at all? This case was the situation of the Elder Research. Elder Research is part of the Inshango Data Science Fellowship. The mandate of this project is to monitor equipment performance and failure detection to avert hydroelectric power plant equipment failure by better scheduling equipment maintenance.
In the absence of labelled data for use, the team relied on unsupervised learning for anomaly detection. The team built an isolation forest, a One-Class Support Vector Machine (SVM) model for multivariate time series. It used t-distributed stochastic neighbourhood embedding (t-SNE) and Principal Component Analysis (PCA) to visualize the output. To improve the detection of anomalies, the team worked on an Unsupervised Deep Anomaly Detection for Multi-Sensor Time-Series Signals model to enhance the detection of abnormalities. Additionally, the team employed time series hydroelectric multisensory data collected by Supervisory Control and Data Acquisition (SCADA). They analyzed the data using time-dependent anomaly detection methods to identify failures. Cyrille Feudjio and Faith Neema Benson were the AIMS alumni who worked on this unique project. We focus on Cyrille for this edition and bring you Faith in the next edition.
Cyrille Feudjio is a Cameroonian national. He attended AIMS Cameroon Centre and graduated in 2021 with a Masters in Mathematical Sciences specializing in Industrial Mathematics and Data Science. He took the Cooperative Master option and wrote his thesis on “The Predictive Effect of Feature Extraction Methods on EEG Dataset: Case of An Epileptic Study.” From November 2014 to December 2017, he attended the National Advanced Polytechnic School of Douala (ex-Faculty of Industrial Engineering). He graduated with a Master of Engineering in Industrial Engineering (Specialization in Industrial Design Technology). He wrote a thesis on “Revamping of a barge to transport the populations of Douala to Manoka.” He was at the University of Dschang, Cameroon, for his Master 1 in Physics from October 2010 to October 2014.
Cyrille started working on an Elder Research project in September 2021. He supported Elder Research to develop a model for detecting anomalies in hydroelectric power equipment schedule maintenance to avert electricity generation. As an industrial Engineer and Industrial Mathematician, AIMS-Ishango Data Science Fellowship Program offered exposure and a chance to work for an international firm, acquire knowledge from experienced professionals, and improve job-clinching chances. He said, “Gain real-world experience on Data Science projects from top companies from around the world, seek guidance from experienced mentors and learn from them, learn new skills and improve the existing experiences to become a senior Data Scientist. Besides, I wanted also to increase my network and gain exposure to global data science opportunities. After completing this fellowship program, I would like to get a job as a data scientist and then use my various skills to add value to the company while developing models that can make positive change in humanity.”
In May 2021, Cyrille joined Philjohn Technologies, a Douala-based IT company, as the Lead Developer. He leads a pool of developers, assists the digital content manager in elaborating technical specifications, guides and supports customers and developers in deploying new websites and developing templates. From June 2021 to April 2021, Cyrille was an Analytics and App Consultant at Philjohn Technologies. He developed sites with functionalities such as membership, chats, and fundraising. He also developed interactive dashboards for processing, analyzing, and visualization using Power BI.
Cyrille interned as a Structural Design Engineer in several Engineering companies, including Cameroon Shipyard and Industrial Engineering (Jul 2017 – Sep 2017), All Industrial Maintenance Service (Apr 2018 – Jun 2018), Global Technology Service (Jul 2018 – Sep 2018), Tawa Engineering (Mar 2019 – Apr 2019), Multilogistic & Services (Jul 2019 – Aug 2019).