Predictive Modeling of Public Sentiment Toward MONUSCO's Mandate: A Machine Learning Approach

Abstract

In eastern Democratic Republic of the Congo, persistent insecurity and protracted armed conflict have intensified public dissatisfaction with the United Nations Organization Stabilization Mission in the Democratic Republic of the Congo (MONUSCO). In Butembo, debates over renewing MONUSCO's mandate have become increasingly contentious, yet empirical evidence on the determinants of public opposition remains limited. This study addresses this gap by using machine learning to predict refusal to extend MONUSCO's mandate from primary survey data. A cross-sectional quantitative survey was conducted among 5518 respondents in Butembo using a structured questionnaire comprising 28 items on sociodemographic characteristics and perceptions of MONUSCO's performance and local security conditions. After excluding the outcome variable, exploratory factor analysis identified latent structures in the data. Fifteen variables with factor loadings above 0.40 were retained and grouped into two dimensions: sociodemographic factors and self-conviction factors. These variables were then used to train and test three classifiers: decision tree, logistic regression, and support vector machine (SVM). SVM yielded the best predictive performance, achieving 93.1% accuracy on the held-out test set, compared with 90.8% for logistic regression and 89.6% for decision tree. The best-performing model was further deployed in a Flaskbased web prototype for real-time prediction. Overall, the study demonstrates the value of combining latent variable modeling and machine learning to analyze public opinion in conflict-affected settings.

Received: 26 September 2025 | Revised: 3 March 2026 | Accepted: 24 March 2026

Conflicts of Interest

The authors declare that they have no conflicts of interest to this work.

Data Availability Statement

The data that support the findings of this study are openly available in Kaggle at https://www.kaggle.com/datasets/staniherstaniher/monusco-dataset-butembo

Author Contribution Statement

Nsenge Mpia Héritier: Conceptualization, Methodology, Software, Validation, Formal analysis, Investigation, Resources, Data curation, Writing – original draft, Writing – review & editing, Visualization, Supervision, Project administration. Kambale Kasambya Moïse: Methodology, Software, Validation, Formal analysis, Investigation, Resources, Supervision.