blaise-mwangi/kenya-mental-health-adolescents

A project analyzing adolescent mental health trends in Kenya. Mental Health Modeling for Kenyan Youth Business Problem NGOs and development partners are increasingly investing in adolescent mental health in Kenya but often lack the data-driven insights needed to guide their support effectively. YouthMind Care, a youth-focused mental health provider, partners with schools to deliver therapy and well-being programs. However, with limited therapists and growing demand, identifying which students need urgent intervention remains a challenge—especially in schools without on-site mental health professionals. Although screening tools are in use, they offer limited capacity to interpret complex psychosocial and demographic patterns or to prioritize follow-up efficiently. To address this, you have been delegated the task of building a data-driven solution that can uncover key risk factors and help allocate mental health resources. Data The dataset used in this project was collected by the Shamiri Institute, an organization focused on improving youth mental health in Sub-Saharan Africa. The dataset includes self-reported survey responses from 2,192 Kenyan adolescents attending secondary schools in Nairobi and Kiambu counties. Project Objectives Which psychosocial and demographic factors contribute to poor mental health outcomes in adolescents? Can we build accurate models to predict students at risk of anxiety or depression using available data? Which machine learning model performs best for predicting mental health risk, balancing accuracy and interpretability? Which specific factors are most strongly linked to high levels of anxiety (GAD-7) and depression (PHQ-9)? Conclusions Rationale Complex Interactions: The relationship between mental health and factors like social support, academic self-perception, loneliness, and demographics is non-linear and interdependent. ML models can capture these interactions better than simpler statistical summaries. Model Choice: Logistic Regression was selected as the top-performing model …