Short-Term Reactive Power Forecasting Based on Real Power Demand Using Holt-Winters’ Model Ensemble by Global Flower Pollination Algorithm for Microgrid

Table 9

XGBoost parameters for regression analysis.

Cluster level	Parameters	Range	SHAP	Associate rule
Sample-1	Objective	Squared error	—	—
	L-Rate	0.032	0.04	0.11
	Max.Depth	2.86	13	0.11
	Bytree	0.51	0.7	0.28
Sample-2	Objective	Squared error	—	—
	L-Rate	0.067	0.048	0.12
	Max.Depth	3.34	14.15	0.12
	Bytree	0.58	0.80	0.31
Sample-3	Objective	Squared error	—	—
	L-Rate	0.070	0.053	0.14
	Max.Depth	3.80	14.99	0.12
	Bytree	0.66	0.90	0.34