Jahangirnagar University, Savar, Dhaka, Bangladesh

Jahangirnagar University, Savar, Dhaka, Bangladesh

Syed Ahmmed College, Rajshahi, Bangladesh

Sukhan Pukur High School, Rajshahi, Bangladesh

February 01, 2025 to Present

February 01, 2020 to January 31, 2025

February 01, 2014 to January 31, 2020

February 01, 2012 to January 31, 2014

Multi-component Weather Prediction using Self-Supervised Multi-Fidelity Ensemble Learning with Model Explainability.

Explainable Ensemble Learning Framework for Accurate Detection of Automobile Insurance Fraud.

Assessment of Socioeconomic and Demographic Risk Factors for Low Birth Weight using Model-agnostic Explainable Ensembles.

A Comparative Study of GARCH and Deep Learning Models in Predicting Bitcoin Daily Returns.  

An Explainable Machine Learning-Based Employee Attrition Predictive System.

Changes in the Board of Directors’ Number of Meetings: why and so what?

Identification of Bacterial Key Genera Associated with Breast Cancer using Machine Learning Techniques.

Boosting Heart Attack Prediction Performance: An Ensemble Learning Perspective. Journal of Science and Technology.

Phishing Website Identification: Unleashing The Potential of Machine Learning and Stacking Ensembles Techniques.

Forecasting Monthly Export of Readymade Garments by Removing Seasonal Impact.

Identifying the Socioeconomic and Demographic factors affecting the Maternal health care and delivery types of Santal women’s of Dinajpur, Bangladesh.

A study on the effects of different factors on Academic achievement among university students in Dinajpur District, Bangladesh: A Statistical Study.

Socio-Economic and Demographic Factors Influencing Fertility Preference in Bangladesh: Evidence from BDHS 2007-2018.

Forecasting the Remittance Inflow Based on Time Series Model in Bangladesh.

Forecasting the Production of Jute Based on Time Series Model in Bangladesh.

Forecasting the Production of Sugar Cane Based on Time Series Model in Bangladesh.

Factors Influencing Women's Waiting Time to First Birth in Bangladesh: An Application Of Cox Proportional Hazard Model.

Time Series Modeling and Forecasting of CPI of Bangladesh.

Prevalence of comprehensive knowledge about HIV/AIDS among ever married men and women in Bangladesh.

Socioeconomic and Demographic Determinants: Malnutrition of 6-59 months old rural santal children and Food security status of their families in Dinajpur.

Predicting Mpox Outbreaks Using Machine Learning, Deep Learning, and Explainable AI for Public Health Interventions. 

8th INTERNATIONAL CONFERENCE ON THE ROLE OF STATISTICS AND DATA SCIENCE IN 4IR (ICRSDS4IR) Department of Statistics, University of Rajshahi, Bangladesh,   December 26 – 28, 2024.

Optimizing Food Cart Revenue Prediction Using Machine Learning and Explainable AI Techniques.  

8th INTERNATIONAL CONFERENCE ON THE ROLE OF STATISTICS AND DATA SCIENCE IN 4IR (ICRSDS4IR) Department of Statistics, University of Rajshahi, Bangladesh,  December 26 – 28, 2024.

Prediction of Wind Speed Using Real Data: An analysis of Statistical Machine Learning Techniques.

Forecasting Day-ahead Solar Radiation Using Machine Learning Approach.

Funded by: HSTU

Position: Associate Professor

Description: At post-graduate and undergraduate’s level, Department of Statistics, Faculty of Science, Hajee Mohammad Danesh Science and Technology University, Bangladesh.

Funded by: Institute of Research and Training (IRT), HSTU, 2024-25.

Position: Principal Investigator

Description: This research addresses the critical public health challenge of low birth weight (LBW) by creating an intelligent and interpretable prediction system. Leveraging Bangladesh Demographic and Health Survey (BDHS) data, we first identify significant socioeconomic and demographic determinants of LBW. We then engineer a sophisticated stacking ensemble model, SmartFusion-LR5, which outperforms standard machine learning and deep learning models with 93% accuracy and a 94% AUC. Beyond raw predictive power, we integrate Explainable AI (XAI) techniques to ensure transparency, revealing the global and local impact of features like maternal age and household wealth index. The final framework offers a practical, scalable solution for early LBW risk assessment, enabling proactive, data-informed healthcare strategies to improve neonatal outcomes in Bangladesh and similar regions.

Funded by: Institute of Research and Training (IRT), HSTU, 2023-24.

Position: Principal Investigator

Description: This project develops and validates a dynamic prediction framework to assess the prognostic accuracy of time-varying biomarkers for patients with Primary Biliary Cholangitis (PBC). Utilizing a landmarking approach and the R package dynamicLM, we model the 5-year risk of clinical events while accounting for competing risks. The research evaluates scenarios involving both baseline covariates with time-varying effects and truly time-dependent covariates. Our analysis demonstrates that the Cause-Specific Cox (CSC) model consistently outperforms a null model across landmark times of 0 to 3 years, showing superior predictive accuracy (lower Brier Scores) and discriminative ability (higher AUC). However, calibration plots reveal a tendency to overestimate risk in mid-ranges and underestimate it for high-risk patients, indicating a need for model recalibration for clinical use. This work underscores the critical value of dynamic prediction models that incorporate evolving patient data, providing a more precise tool for guiding timely interventions and improving long-term patient outcomes in chronic diseases.

Funded by: Institute of Research and Training (IRT), HSTU, 2022-23.

Position: Principal Investigator

Description: This project addresses the critical business challenge of employee churn by developing a machine learning solution to predict staff attrition. The core objective was to identify the key factors influencing an employee's decision to leave and to build a predictive model that can flag at-risk individuals. Using the IBM analytics dataset, several classification models—including Logistic Regression, Random Forest, and Lasso Regression Classifier—were trained and evaluated. A comparative analysis revealed that the Lasso Regression Classifier was the most effective overall, achieving the highest scores in Accuracy, ROC Accuracy, and Recall, while also producing the lowest rate of false-negatives (10.27%). The Random Forest model, however, achieved the highest Precision. The analysis also identified key employee attributes correlated with a higher risk of attrition. These include being young, having a low income, working overtime, being single, and holding roles such as Lab Technician or Sales Representative. This predictive system enables organizations to proactively identify employees who are likely to leave. By understanding these risks and the underlying factors, companies can implement targeted retention strategies, ultimately reducing turnover costs and retaining essential talent.

Funded by: World Bank (HEQEP) and UGC, Tk. 3,00000.00 USD $2710.60

Position: Member, SUB-PROJECT MANAGEMENT TEAM (SPMT)

Description: From 2014 to 2017, our research team adeptly implemented an initiative aimed at enhancing academic quality on an international scale, which was financially supported by the Higher Education Quality Enhancement Project (HEQEP) through the Academic Innovation Fund (AIF). This all-encompassing endeavor sought to modernize both the academic and physical infrastructure of the department. Notable achievements by the team encompassed a significant refinement of the Statistical Data Mining curriculum, which was specifically designed to align with avant-garde applications in the fields of Bioinformatics, Climatology, and Health. This academic enhancement was further bolstered by considerable upgrades to the physical facilities, including enhancements to laboratories, a workshop area, restroom facilities, a seminar library, and a computer center. In order to ensure a sustainable impact, the initiative additionally prioritized capacity building through the elevation of both academic and non-academic staff quality. Our initiatives comprised the establishment of an inter-departmental committee aimed at fostering collaboration, the enhancement of the scientific infrastructure, and the development of a more contemporary and secure departmental website, ultimately resulting in a comprehensive transformation of the department.

Funded by: Institute of Research & Training, HSTU, 2014-15.

Position: Principal Investigator

Description: This project focuses on forecasting the agriculture and forestry sector's contribution to Bangladesh's gross domestic product (GDP) using time series analysis. While Auto-Regressive Integrated Moving Average (ARIMA) models have been widely applied in financial market forecasting for decades, their use in agriculture and forestry GDP forecasting has been limited. The study analyzed published secondary data for Bangladesh’s agriculture and forestry sector from 1979-80 to 2012-13. Through the Box-Jenkins methodology, the ARIMA (0, 2, 1) model was identified as the best-fit model based on several statistical criteria such as adjusted R-squared, Akaike Information Criterion (AIC), Bayesian Information Criterion (BIC), and Mean Absolute Percentage Error (MAPE). The model’s forecasts closely matched the original data series, indicating accuracy and reliability for short-term forecasting. This validated model can predict agricultural and forestry production trends effectively, both during and beyond the estimation period. The study’s findings are valuable for policymakers as they provide a robust quantitative tool to support decision-making in the agriculture and forestry sectors, enhancing strategic planning and sustainable development efforts in Bangladesh.

URL: https://scholar.google.com/citations?user=08pZ5psAAAAJ&hl=en&oi=ao

URL: https://www.researchgate.net/profile/Md-Hassan-61?ev=hdr_xprf