Associative rule mining is defined as the task that deals with the extraction of hidden knowledge and frequent patterns from very large databases. Traditional associative mining processes are iterative, time consuming and storage expensive. To solve these processes, a way of representation that reduces this size and at the same time maintains all the important and relevant data needed to extract the desired knowledge from transaction databases is needed. This paper proposes a method that merges the transactions in the transaction database and uses FP-Growth algorithm for mining associative knowledge is presented. The experimental results in terms of compression ratio, both in terms of storage required and number of transactions, prove that the proposed algorithm is an improved version to the existing systems.