Development of predictive modeling for suicidal ideation using analytic hierarchy process and data mining approaches in Malaysia

• Main Author: Chan, Sin Yin
• Format: Thesis
• Language: English; English
• Published: 2025
• Subjects: HN Social history and conditions. Social problems. Social reform
• Online Access: https://etd.uum.edu.my/11956/1/s829055_01.pdf; https://etd.uum.edu.my/11956/2/s829055_02.pdf; https://etd.uum.edu.my/11956/

Suicide is a global public health concern, claiming over 720,000 lives annually and standing as the second leading cause of death among young adults. In Malaysia, there are approximately 250 Malaysians attempt suicide each day and at least three fatalities are recorded. Nevertheless, only few studies have explored the factors driving suicidal ideation in young adults or predicted its likelihood. This study seeks to identify and rank the key factors influencing suicidal ideation among Malaysian young adults. Subsequently, the individuals are segregated into distinct groups based on their risk profiles. Finally, predictive models are developed to anticipate the likelihood of experiencing suicidal ideation. This study employed two sets of questionnaires. The first questionnaire was used to rank the influential factors using Analytic Hierarchy Process whereas the second questionnaire was designed to develop a predictive model for assessing the tendency toward suicidal ideation. Stratified sampling was first implemented to determine the sample size for each state and the respondents were then selected randomly from each state. A total of 891 valid responses were obtained. The findings revealed that previous suicide attempts, negative life events, and financial problems were the most influential factors for suicidal ideation. Young adults were successfully segregated into three distinct clusters, representing lower, moderate and higher risk groups using cluster analysis, providing valuable insights into population characteristics. Results showed that Decision Tree using Gini index with three branches outperformed the other predictive models by achieving the lowest misclassification rate. Integrating decision-making frameworks and data mining supports evidence-based decisions that strengthen public health policies and assist healthcare professionals in timely intervention. The findings also contribute to protecting vulnerable populations while promoting societal harmony and public health which aligns with Malaysia’s National Health Policy