This document presents a study on using a custom-built Apriori algorithm for web mining and discovering frequent patterns in web log data. The key steps are: (1) preprocessing a 70MB web log file, (2) developing a custom Apriori algorithm that prunes candidate itemsets of size >2 and only considers transactions of size >=k, (3) identifying frequent patterns using the custom algorithm, (4) analyzing the discovered patterns, and (5) developing a software tool to implement the custom algorithm. Experiments show the custom algorithm takes less time than the classical Apriori algorithm. The study aims to efficiently mine useful knowledge from web data.