Finding frequent itemsets remains challenging due to manual threshold specification requirements in existing algorithms. This paper presents an Enhanced FP-Growth algorithm incorporating a hybrid adaptive support threshold that combines statistical variance analysis, frequency distribution patterns, and transaction density metrics. The algorithm automatically adjusts support levels based on dataset characteristics, eliminating manual threshold tuning. Experimental evaluation on five benchmark datasets against Aprior, FP-growth, and FP-Max shows our Enhanced FP-Growth consistently achieves superior execution time and improved memory efficiency. The hybrid threshold mechanism dynamically calibrates according to dataset characteristics, offering substantial efficiency gains across diverse data types.