Year: 2025
Type: article
Abstract: The integration of Machine Learning into Business Intelligence represents a paradigm shift from descriptive to predictive and prescriptive analytics, with Automated Machine Learning (AutoML) democratizing advanced analytics by automating model selection, hyperparameter tuning, and deployment. AutoML platforms bridge the talent gap in data science, enabling business analysts and domain experts to develop sophisticated models without extensive programming expertise. These platforms employ various architectural approaches, including sequential optimization, parallel search, and neural architecture search, extending capabilities across the entire analytics workflow from data preprocessing to model deployment. Performance evaluations across forecasting, classification, and anomaly detection tasks reveal that AutoML solutions achieve competitive accuracy compared to manually tuned approaches for many business applications, with particular strengths in multivariate forecasting and handling imbalanced classification problems. Despite promising performance, integration into existing enterprise BI ecosystems presents technical challenges, including data pipeline compatibility, computational resource requirements, and real-time scoring latency, alongside organizational hurdles related to model transparency, skills transition, and governance. A four-phase implementation framework offers guidance for successful AutoML adoption, emphasizing capability assessment, platform selection, pilot implementation, and scaled deployment with continuous improvement.
Author: Aravind Puppala
Language: English
Subfield: Artificial Intelligence
Field: Computer Science
Domain: Physical Sciences
Open Access status: hybrid
OpenAlex ID: W4411310178