Comparison of classification algorithms for predicting autistic spectrum disorder using WEKA modeler

Prior research has shown that integrating large datasets with automated learning tools enhances diagnostic precision in clinical settings. Early identification of developmental conditions remains a priority for ensuring timely therapeutic interventions for affected individuals. That uncertainty drove the need for more robust screening protocols to mitigate human error during patient assessments. No prior work had resolved the specific challenges posed by incomplete medical records in this diagnostic domain. Existing literature often overlooks how absent data points influence the reliability of predictive models. This gap motivated a closer examination of how data preprocessing affects algorithmic performance. Previous investigations frequently prioritized model selection while neglecting the integrity of the underlying input information. Addressing these limitations provides a clearer path toward reliable automated screening for neurodevelopmental conditions.

The aim of this study is to evaluate the performance of six top-tier classifiers for predicting Autistic Spectrum Disorder occurrences. Researchers sought to determine which algorithm provides the highest diagnostic accuracy when analyzing screening datasets. This investigation addresses the persistent challenge of human error in clinical assessments by utilizing automated computational tools. The team specifically examined how missing information within patient records affects the reliability of these predictive models. By implementing a systematic imputation method, they aimed to improve the quality of the input data. The study also explores the comparative effectiveness of various algorithms, including J48 and Support Vector Machine. This work is motivated by the need to provide health practitioners with more precise diagnostic aids. Ultimately, the researchers intend to facilitate earlier and more effective treatment plans for patients through improved screening technology.

Review Approach involved evaluating six distinct machine learning classifiers using the WEKA modeler software. The researchers processed an ASD screening dataset to compare predictive performance across several key metrics. They specifically focused on sensitivity, specificity, accuracy, receiver operating characteristic, area under the curve, and runtime. To handle incomplete information, the team applied an imputation method that replaced absent values with the mean of available records. This systematic approach allowed for a direct comparison of model behavior under both clean and original data conditions. The investigation included Naive Bayes, Logistic Regression, K-Nearest Neighbors, J48, Random Forest, and Support Vector Machine algorithms. Each model underwent rigorous testing to determine its efficacy in identifying disorder occurrences. This methodology ensured that the final recommendations were based on consistent and reproducible computational practices.

Key Findings From the Literature indicate that the J48 algorithm consistently produced the most promising results across all tested conditions. The study demonstrates that this classifier maintains high performance even when datasets contain missing entries. In contrast, the Support Vector Machine model did not perform well when applied to smaller, simpler datasets. The researchers observed that cleaning data through mean imputation significantly influenced the final classification outcomes. By addressing absent records, the team improved the reliability of the predictive models compared to previous works. The analysis confirmed that specific algorithmic choices directly impact the accuracy of diagnostic predictions. These results highlight a clear hierarchy in model effectiveness for ASD screening tasks. The data suggests that selecting the right classifier is vital for minimizing errors in clinical diagnostic environments.

Synthesis and Implications suggest that the J48 algorithm consistently outperforms other tested models in predicting ASD. The authors propose that replacing absent data with average values improves the stability of classification results. Their analysis indicates that Support Vector Machine (SVM) models may struggle when applied to smaller, less complex datasets. These findings highlight the importance of rigorous data preparation before applying predictive analytics in healthcare. The researchers emphasize that accurate diagnostic tools can significantly improve long-term outcomes for patients. They suggest that practitioners should prioritize algorithms that demonstrate high sensitivity and specificity in clinical trials. This review implies that automated screening tools serve as valuable supplements to traditional diagnostic methods. The authors conclude that refining these computational approaches will ultimately support more precise clinical decision-making.