ndeyekwade-cmd/High-Dimensional-Data-Analysis-AIMS

Analysis of Digital Habits and Wellbeing using PCA & Multiple Regression - AIMS Senegal Academic Project # High Dimensional Data Analysis - Digital Habits and Wellbeing **AIMS Senegal Academic Project** *Master's in Mathematical Sciences with Big Data Specialization* --- ## 📋 Project Overview This project analyzes the **relationship between digital device usage, social media engagement, physical activity, and psychological wellbeing** among AIMS students using advanced multivariate statistical methods including PCA and Multiple Regression Analysis. ### Team Members - Group 5 - **Ndeye Khady Wade** - **Lieumo NJOPNU Marcelle Alida** - **Lekounda Ngolo Mardochet Gedeon** ### Supervisor Prof. Sophie Dabo (University of Lille) **Date:** January 22, 2026 --- ## 🎯 Research Objectives ### Main Research Questions 1. What are the main patterns in student digital behavior? 2. What factors predict smartphone usage? 3. Does happiness influence screen time? 4. How do digital habits and wellbeing interconnect? --- ## 📊 Dataset Description ### Sample Characteristics - **Sample Size:** 368 AIMS students - **Data Type:** Self-reported survey data - **Missing Values:** 0.27% - **Final Sample:** 368 complete observations ### Variables Collected | Variable | Description | Type | |----------|-------------|------| | `Phone` | Daily phone screen time (minutes) | Continuous | | `SocialNetworks` | Time on social media (minutes/day) | Continuous | | `Happiness` | Happiness score (1-10 scale) | Discrete | | `Walk` | Daily walking distance (km) | Continuous | | `InstagramRatio` | Followers / (Following + 1) | Continuous | --- ## 🔬 Methodology ### Part 1: Multivariate Analysis Techniques #### 1. Singular Value Decomposition (SVD) **Purpose:** Matrix factorization for dimensionality reduction **Applications:** - Image compression - Recommendation systems (Netflix, Spotify) - Noise reduction - Backbone of PCA, CA, and MCA #### 2. Correspondence Analysis (CA) **Purpose:** Analyze contingency tables (categorical data) **Key Features:** - Uses χ² distance - Simultaneous row-column …