There have been significant advancements in artificial intelligence and machine learning technologies which have dramatically changed how Predictive Analytics and intelligent decision-making systems work within numerous sectors. However, traditional machine learning processes involve a lot of manual input from humans throughout various stages of data preprocessing, feature extraction/selection, model choice, and hyper-parameter optimization. As such, the entire workflow is labor-intensive, time consuming and heavily reliant upon specific domain knowledge or expertise. This study presents an autonomous system for automated dataprocessing/preprocessing, model selection, and predictive analysis utilizing novel Automated Machine Learning (automl) techniques and intelligent optimization methods. The proposed autonomous system will include an array of automated preprocessing operations such as: missing value imputation; outlier detection; feature scaling; feature engineering; intelligent data transformation with an adaptive strategy for selecting appropriate models and their associated hyperparameters. To optimize both predictive accuracy and computational efficiency, this system utilizes a variety of optimization techniques including: bayesian optimization; Hyperband tuning; reinforcement learning-based optimization; and automl methodologies. In addition, several different machine learning models, including Random Forest; XGBoost; LightGBM; and AutoGluon are tested to determine the most accurate predictive models. Findings from experimental results demonstrate the proposed autonomous system is significantly better than typical machine learning workflows with respect to efficiency of preprocessing; predictive performance; scalability; and reliability of decisions made by Intelligent Systems. Furthermore, this study highlights the significance of developing intelligent autonomous systems to automate end-to-end Predictive Analytics applications across a range of sectors including health care forecasting; financial prediction; education analytics; industrial automation; and smart systems. Finally, this research contributes towards the development of next generation autonomous artificial intelligence systems capable of reducing human involvement and increasing predictive intelligence through fully Automated Machine Learning workflows.

 Keywords:

Automated Machine Learning, Predictive Analytics, Intelligent Systems, Bayesian Optimization, Reinforcement Learning, data preprocessing, automl, artificial intelligence