MetricPredictor enables predictive autoscaling by analyzing historical metrics and generating forecasts for future resource demand.

Configure data sources to collect metrics from KEDA resources (ScaledObjects/ScaledJobs) or CSV files. The predictor will analyze patterns in your data to make accurate predictions.

Set up model parameters including prediction algorithms (Prophet, Linear, ARIMA), history length for training data, and sample rates for data collection.

Advanced model configuration allows you to account for holidays and seasonal patterns, improving prediction accuracy for workloads with time-based variations.