Raphael BRARD

– Contributed to the development of a predictive algorithm to estimate hiking trail usage based on GPS data from Outdoorvision and Strava Metro for the Écrins National Park.

– Developed an algorithm to identify potential future premium clients. – Analyzed customer purchasing behavior to support digital marketing for the loyalty platform.

– Led requirements elicitation and defined a structured action plan. – Built a centralized database integrating company data, partner sources, and open APIs (Météo-France, Hubeau). – Developed a predictive tool estimating canoe traffic on the Ardèche River at D+10 using environmental and tourism indicators. – Designed a validation protocol comparing Outdoorvision GPS data with Eco-Compteur trail sensors. – Established a methodology to estimate real territorial attendance from GPS traces. – Conducted statistical analyses to assess the impact of running and hiking in protected areas, supporting biodiversity protection strategies.

– Performed statistical analyses of longitudinal proteomics and RNA-seq datasets. – Conducted multiple analyses following protocols developed with medical experts in the context of an early-stage drug development study. – Collaborated on a public-private 3TR project with Kiel University: designed a statistical protocol for analyzing a longitudinal cohort using adapted clustering methods, and analyzed cluster patterns. – Authored multiple scientific articles and conference posters.

– Redaction and publication of a scientific article about the develop Walk Detection Pipeline in the Sensors journal. – Presentations of a posters at the CMStatistics 2021 conference about our pipeline. – Comparaison of the internal walk analysis algorithms eGait to the exisiting state-of-art solution GAITRite in the CHU of Nantes.

– Development of walk detection pipeline from data record in a real-life environment via an IMU sensor wear on the right hip. – Data Management and Feature engineering of non-euclidean multivariate time series of dimension 4 (quaternions time series). – Tunning and comparaisons of the results of different ML model include in my pipeline. – Implementation of the algorithms inside a R package stored in GitLab. – Performed a literature review of methods used in Human Activity Recognition for walk detection based on IMU sensor data. – Presented results to non-statistician partner teams involved in the project.