Murtaza M. Junaid Farooque

Machine learning models for ultra-high-performance concrete (UHPC) commonly report coefficients of determination above 0.94 on held-out records, and these values now shape expectations for data-driven mix design. We show that a substantial part of this reported predictive performance depends on how the models are evaluated. Benchmark UHPC datasets contain replicate records, meaning the same mixture tested at several curing ages or specimen sizes. Random train-test splitting distributes these replicates across both sides of the split, so models are scored partly on mixtures encountered during training. Using an open dataset of 1,193 UHPC records and an independent benchmark of 1,030 conventional concrete records, we compared random cross-validation with mixture-grouped cross-validation for eight regressors and four target properties. Under mixture-grouped evaluation the coefficient of determination for compressive strength fell from 0.947 to 0.893 and for flexural strength from 0.897 to 0.779, while root-mean-square error rose by 42 to 47% relative to the value implied by random splitting. A controlled experiment in which sample size was held constant and only the number of records per mixture varied reproduced the effect on both datasets, with the gap growing from approximately zero at one record per mixture to 0.13 at four. Conformal prediction intervals calibrated on randomly selected records covered new mixtures 82 to 84% of the time against a 90% nominal level, whereas calibration on held-out mixtures restored approximately nominal coverage with wider intervals. We provide an evaluation and reporting protocol for concrete machine learning.

This study investigates the effectiveness of predictive performance of multiple modeling techniques in forecasting Nigeria’s Food Price Index (FPI) using monthly data from the Food and Agriculture Organization (FAO) Food Price Index spanning January 1990 to January 2025. Five major food subsector indices—Meat, Dairy, Cereals, Oils,and Sugar—serve as predictors. Initial regression on the raw series produced inflated R2 values, prompting diag‑nostic tests that revealed severe non‑stationarity and multicollinearity. These issues were addressed through a two‑step transformation: natural logarithmic conversion to stabilize variance, followed by first‑order differencing to achieve stationarity and eliminate false trends. Post‑transformation diagnostics confirmed full correction, with all variables stationary and multicollinearity reduced to acceptable levels. Using standardized log‑differenced data, five models were evaluated: Multiple Linear Regression, Artificial Neural Network, Random Forest, Support Vector Machine, and Long Short‑Term Memory (LSTM). On the transformed scale, the Support Vector Machine was the strongest performer; however, after back‑transforming predictions to the original scale through recursive exponential reconstruction, Random Forest achieved the highest accuracy (R2 = 0.9342, MAPE = 0.65%). LSTM models performed poorly, indicating a structural mismatch with differenced series lacking persistent trends. The study provides three methodological insights: (1) top performance on transformed data does not guarantee accuracy after inverse conversion; (2) log‑differencing is highly effective for resolving non‑stationarity, heteroscedasticity, and multicollinearity while preserving economic interpretability; and (3) ensemble averaging enhances recursive multi‑step forecasting stability. For practical policy applications requiring multi‑month food price projections, Random Forest is recommended, with Multiple Linear Regression serving as a transparent alternative for coefficient interpretation. These findings support the design of more informed food security interventions and market stabilization strategies in Nigeria.

The use of algorithmic decision systems is being expanded to high-risk areas like credit, recruiting, and distributing government resources. Despite the fact that these systems are usually claimed to be objective and efficient, there have been apprehensions about the likelihood of structural inequalities being perpetuated by the systems. This paper examines the effect of a fairness-aware pre-processing technique called reweighing on the performance of a predictive system in a controlled simulation environment. Using a synthetically created credit approval dataset with structural disadvantage embedded, we compare the performance of a logistic regression classifier with and without reweighing. Fairness is measured using demographic parity disparity (DPD), disparate impact ratio (DIR), and equalized odds difference (EO), along with predictive accuracy. In a single test scenario (seed = 42), reweighing does not improve all fairness metrics uniformly. However, when analyzed for robustness across 50 independent random seeds, we find modest average reductions in demographic parity disparity and equalized odds difference for reweighing, with little change in predictive accuracy. Threshold sensitivity analysis also shows that fairness metrics are sensitive to decision thresholds. These results show that fairness-aware pre-processing can lead to systematic improvements in expectation, although trade-offs across fairness metrics and performance remain context-dependent.

In medicine, it is well known that healthy individuals have different physical and mental characteristics. Ancient Indian medicine, Ayurveda and the Persian-Arabic traditional Unani medicine has two distinct approaches for the classification of human subjects according to their temperaments. The individual temperament is an important foundation for personalized medicine, which can help in the prevention and treatment of many diseases including COVID-19. This paper attempts to explore the relationship of the utmost important concepts of these systems called individual temperament named as Prakruti in Ayurveda and Mizaj in Unani practice using mathematical modelling. The results of mathematical modelling can be adopted expediently for the development of algorithms that can be applied in medical informatics. For this, a significant literature review has been carried out. Based on the previous researchers' reviews the essential parameters have been identified for making the relationship and hypothesis were framed. The mathematical modelling was adopted to propose the existence of the relationship between the parameters of such an ancient and rich medicine systems. The hypotheses are validated through the mathematic driven model. Doi: 10.28991/esj-2021-01258 Full Text: PDF

Data Mining Application in Classification Scheme of Human Subjects According to Ayurvedic Prakruti – Temperament