Enhancing Public Cloud Performance Forecasting with Advanced ML Techniques
Kamparapu Nagini Saraswathi Ishwarya, Gandhikota Umamahesh · 2025
Cloud computing administrations are essential for tending to the developing requirement for process escalated applications by offering efficient computational and stockpiling assets. With the rising reliance on cloud administrations, it is fundamental to further develop cloud asset designation. CloudProphet presents an original machine learning procedure for estimating the exhibition of “virtual machines (VMs)” in cloud settings. The methodology uses “Dynamic Time Warping (DTW)” to arrange application sorts and utilizes Pearson relationship to find emphatically related runtime measurements. The estimations are used in three variations of machine learning algorithms: LSTM without profoundly chose and DTW measurements, LSTM with exceptionally chose and DTW measurements, and GRU with DTW and exceptionally connected measurements. The GRU model, coordinating both DTW and profoundly corresponding markers, outperforms others, achieving 99.30% precision in gauging VM execution. The philosophy is substantiated using a cloud dataset from GitHub and increased by means of continuous trial and error with live datasets. This shows its viability in exactly gauging both application classifications and VM execution, with the discoveries highlighting the GRU model's transcendence in cloud asset the executives in common sense settings.