SJSakib Salam Jamee

Sakib Salam Jamee

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article


Sakib Salam Jamee has a total of 14 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    WPIS: Machine Learning-Driven E-Commerce Churn Prediction Integrating Technical Metrics and Information Quality for Strategic MIS Decision Support

    by Mainul Islam Khan, Md Abutaher Dewan, Khandakar Rabbi Ahmed, Sakib Salam Jamee, Md Asif Hassan Reon, Md Shohan Bhuiyan, Md Nayem Rahman

    E-commerce website performance—encompassing both technical delivery quality and information quality—is associated with customer retention: platforms exhibiting poor technical metrics (session friction, high cart abandonment) or poor information quality (low engagement, weak content relevance) tend to show elevated churn. This study operationalizes website performance prediction by mapping behavioural and transactional signals to a Technical Metrics sub-vector Ti and an Information Quality sub-vector Qi (theoretically assigned proxy indicators), constructing a Website Performance Influence Score (WPIS), and training an XGBoost prediction engine that achieves 92.1±0.3% accuracy across five stratified train–test splits, with all preprocessing parameters fit exclusively on each split’s training fold. Using a publicly available E-Commerce Customer Behavior dataset of 50,000 records, the model attains F1-scores of 0.95 (low-risk) and 0.86 (high-risk) and an AUC-ROC of 0.93 for the churned class. Feature importance, cross-validated against SHAP attributions, identifies customer service interactions, customer lifetime value, discount usage rate, and cart abandonment rate as the dominant performance-failure signals, with the engineered Information Quality Degradation Index (IQDi) and Technical Performance Failure Score (TPFi) ranking sixth and ninth by gain. A Strategic MIS Decision Support Layer stratifies customers into risk tiers and maps each tier to illustrative, scenario-based intervention strategies. These results support the hypothesis that ensemble learning captures non-linear interactions between technical and information-quality dimensions, offering a reproducible, theoretically grounded foundation for data-driven MIS decision-making in digital commerce.

  • Open Access

    Cryptographically Secured Machine Learning for Resilient Multi-Tier Supply Chains

    by Alfaiz Madhiya, Vijay Solanki, Khandakar Rabbi Ahmed, Sakib Salam Jamee, Mainul Islam Khan, S. M. Rezvi

    As we move further into a more digital supply chain, we have seen tremendous improvements in terms of efficiency, but we have also seen a rise in the risks that data vulnerabilities pose, particularly for intermediary supply chain nodes. As a result, traditional security measures have proven inadequate in protecting critical data that flows through these intermediary supply chain nodes, making them more susceptible to possible security breaches and tampering. In this paper, we introduce a novel solution that utilizes cryptography and Artificial Intelligence (AI)-powered predictive modeling for the security of supply chain data. Unlike other solutions that only protect endpoint data, our solution provides a more comprehensive security solution that extends cryptography for intermediary supply chain data, making it more secure and protected from possible breaches and tampering. To further improve our solution, we have utilized AI models, namely, XGBoost, Random Forest, and LightGBM, for predictive modeling. To prevent target leakage, features arithmetically derived from the recovery-duration target were excluded from the predictor set prior to training. The results indicate that, once the leakage-affected features are removed and hyperparameters are properly tuned, all five models achieve modest but genuine predictive accuracy, with R2 values in the range of 0.542–0.552. The best-performing model on this leakage-audited feature set is XGBoost, with an R2 value of 0.5525 and a test RMSE of 38.16 days, closely followed by Linear Regression and Ridge Regression (RMSE = 38.22 days, R2 = 0.5512), with the practical difference between the two being small (0.06 days RMSE) despite being statistically consistent across ten random splits. This demonstrates the solution’s effectiveness in reducing security risks while maintaining realistic, leakage-free predictive accuracy.

  • Open Access

    Detection of False Stealthy Data Injection Attacks in Smart Meters Using Machine Learning and Blockchain Technology

    by Mohiuddin Mehedi, Abdul Aziz Kabir, Khandakar Rabbi Ahmed, Furqaan Mujtahid, Sakib Salam Jamee, Md Nayem Rahman

    Despite the benefits associated with the use of smart meters in advanced metering infrastructure, the widespread deployment of such meters has introduced vulnerabilities that leave power systems susceptible to stealthy false data injection attacks which cannot be detected by conventional methods. In this paper, we present a framework that combines a bidirectional long short-term memory network with an attention mechanism and a blockchain integrity layer to provide secure anomaly detection. We utilize the Smart Meter Electricity Consumption Dataset, augmented with synthetically injected anomalies, to detect abnormal consumption behavior. Experimental results demonstrate that the proposed hybrid model achieves up to 96.23% accuracy, 99.88% precision, 92.58% recall, and 96.09% F1-score, outperforming eXtreme Gradient Boosting (XGBoost), Isolation Forest, and Random Forest baselines. Confusion matrix analysis confirms minimal false positives and strong detection capability for stealthy attacks. The blockchain layer ensures immutability and trustworthiness of detection results through cryptographic hashing and consensus mechanisms with negligible overhead. The proposed framework offers a scalable, secure, and interpretable solution for defending smart grid infrastructures against complex cyberattacks.

  • Open Access

    Attn-ChurnNet: A Transformer-Based Sequential Framework for Customer Churn Prediction in Subscription Platforms with Focal Loss Training and Conformal Uncertainty Quantification

    by Didar Hossain, Mohiuddin Mehedi, Khandakar Rabbi Ahmed, Md Rafiul Mahmud, Mainul Islam Khan, Sakib Salam Jamee

    Customer churn is a critical challenge for subscription-based digital platforms, as customer activity patterns change dynamically over time. Traditional churn modeling methods fail to account for sequential dependencies across customer interaction histories. This paper presents Attn-ChurnNet, a novel attention-based Transformer architecture that effectively predicts customer churn by modeling sequential customer contact histories. The proposed methodology leverages multi-head self-attention with sinusoidal positional encodings and Pre-LN residual connections to highlight key interaction sequences and interpret the temporal dynamics of customer activity in subscription platforms. Experiments are conducted on the large-scale WSDM–KKBox Customer Churn Prediction dataset using a temporal train/validate/test split, incorporating transaction records, usage logs, and customer demographic information. Comprehensive comparison against established baselines—Logistic Regression (LR), Random Forest (RF), XGBoost, and Gated Recurrent Unit (GRU)—demonstrates that Attn-ChurnNet achieves a macro-averaged classification accuracy of 97.03% (±0.41%), precision of 95% (±0.5%), recall of 96% (±0.6%), F1-score of 95.50% (±0.5%), AUC of 0.98 (±0.004), Average Precision of 0.963, and log-loss of 0.15 (±0.007) under five-fold stratified cross-validation, outperforming all competing approaches with statistical significance (p<0.01, McNemar’s test). A comprehensive two-part ablation study (22 variants), calibration analysis (ECE = 0.031; Ts* = 1.08), attention entropy analysis with Jensen–Shannon divergence and two-sample t-test (t=18.4, p<0.001), Integrated Gradients attribution, conformal prediction (91.4% coverage, 88% singleton efficiency), precision–recall analysis, and computational complexity evaluation further validate the model’s design and production readiness.