Big Data query processing mainly deals with executing queries on Big Data. Since, most of the Big Data repositories are made up of unstructured data, coupled with high velocity and volume of data generation, designing efficient query processing techniques imposes significant challenges. Hence, Approximate Query Processing Techniques (AQPTs) are an attractive option. AQPT are ideally suited for executing aggregate queries, where the AQPT provides approximate results with attractive computational efficiency. Recently in the literature, AQPT was presented to execute simple non-join aggregate queries on Big Data. However, this presented AQPT does not deal with the more complex join-aggregate queries. Hence, in this paper, AQPT is presented for the approximate execution of join-aggregate queries. The proposed AQPT is designed using Central Limit Theorem (CLT), and achieves predefined estimation error. An empirical analysis study is presented in which the proposed AQPT is compared against a contemporary technique. In this empirical analysis study, the proposed AQPT significantly outperforms the contemporary technique both in-terms of estimation accuracy and computational latency.