מושגי ליבה
Male victims of domestic abuse are often overlooked, but a study in Bangladesh sheds light on the prevalence, patterns, and factors of male domestic violence.
תקציר
The study explores male domestic violence (MDV) in Bangladesh, highlighting the underexplored realm of male victims. It challenges prevailing notions by using data analysis and machine learning to understand MDV dynamics. The content covers:
Introduction to domestic violence and its forms.
Societal norms in Bangladesh perpetuating male victimization.
Research methodology involving data collection and exploratory data analysis (EDA).
Implementation of machine learning models for predictive modeling.
Integration of eXplainable Artificial Intelligence (XAI) techniques for model interpretability.
Contributions of the study and policy recommendations.
Literature review on DV against men globally.
Detailed methodology including dataset collection, preprocessing, and model development.
Results from EDA showcasing insights into MDV dynamics based on demographic factors.
Statistical tests like Cramer’s V correlation and Chi-square analysis for association between variables.
Model development with traditional classifiers like Logistic Regression, Decision Trees, SVM, Naïve Bayes, Random Forests, K-nearest Neighbors, Gradient Boosting, as well as DL models like Artificial Neural Networks and ensemble models like XGBoost, LightGBM, CatBoost.
סטטיסטיקה
We implemented 11 traditional machine learning models with default and optimized hyperparameters, 2 deep learning models, and 4 ensemble models.
Despite various approaches, CatBoost emerged as the top performer achieving 76% accuracy.
ציטוטים
"Our findings challenge the prevailing notion that domestic abuse primarily affects women."
"ML techniques enhance the analysis and understanding of the data."