This research focuses on developing novel algorithms for two key areas: frequent sequence mining in transactional databases and enhanced load value prediction. A novel algorithm, SPAM (Sequential Pattern Mining Algorithm), is introduced to efficiently discover frequent sequences, even those of considerable length. SPAM leverages advanced pruning and indexing techniques to optimize its search. Furthermore, the research explores load value prediction (LVP) through identifying frequent patterns within program memory access traces. These discovered patterns serve as the foundation for developing efficient pre-fetching strategies, leading to improved performance.