System testing is critical since it assures that users will not encounter faults. Most popular approaches used in software testing are Black-Box and White-Box. Equivalence Partitioning and Boundary Value Analysis are some of the Black Box Testing Method techniques. These approaches are used to minimize the number of test data that could be used in actual testing. Still, if the input range of data is enormous, the problem that arises is a large number of input test data even after using the Equivalence Partitioning and Boundary Analysis. In this work, Machine Learning, Genetic Algorithms, and Decision tree learning techniques are used to select the optimal set of input test data from the extensive range of input, this work aims to reduce the input test cases by using artificial intelligence techniques. The experiments' preprocessed data contain 125,502 records in total. The Genetic Algorithm was run with the following settings: pm = 0.15, Maxgen = 100.000, and pc = 0.7. The results show the classified records accuracy is about 96%, whereas 9 out of 10 selected scenarios are covered. According to priority and criteria, there are ten possibilities in this work that need to be handled. This method demonstrates that nine out of ten cases are addressed, while one is left unclassified. On the other hand, 15 of the 18 attributes are covered as a result of the correlation cut points. These results demonstrated how the method makes use of the most previously identified cases. The biggest flaw in this strategy is that it ignores one of the scenarios created using the classification tree process, which could lead to missing out on some crucial input data. On the other side, the most crucial scenarios (correlation and priority) are addressed concurrently where minimizing the effect of discarding one of the scenarios. Keywords: Genetic Algorithm, Decision Tree Learning, Black Box Testing, Big Data.