Hybrid Machine Learning Based Predictive Model To Improve Cloud Application Performance In Cloud Saas

Cloud computing provides IT resources like applications, infrastructure, and platforms as services over the Internet. Predictive maintenance leverages data-driven methods to anticipate equipment failures, allowing for more efficient maintenance management. This process involves monitoring equipment conditions by collecting data over time. Typically, industrial equipment is used without a clear scheduled maintenance plan. This study presents a new strategy for predictive maintenance in SaaS application cloud infrastructure, focusing on predicting customer churn in Customer Relationship Management (CRM) systems. During the data collection phase, consumer data is extracted from a churn dataset on Kaggle. Preprocessing steps include eliminating null values, scaling features, and converting categorical variables to enhance data quality before analysis. This research evaluates several ML models using recall, F1 score, accuracy, and precision. The findings indicate that the Hybrid SVM + Decision Tree model achieved a 92.71% accuracy rate, outperforming other models and offering more accurate predictions of customer churn in SaaS applications. Future research should explore advanced feature engineering techniques, consider temporal dynamics to improve prediction accuracy, and incorporate real-time data streams for dynamic model adaptation.