Study of an Adaptive Financial Recommendation Algorithm Using Big Data Analysis and User Interest Pattern with Fuzzy K-Means Algorithm

In an ever-changing financial market, big data is set to revolutionize user interest management by sparking innovation and reshaping recommendations for the future. Conventional financial services face significant challenges like accessibility, personalization, limited reachability, and incomplete information about user interest patterns. Thus, it results in suboptimal financial recommendations that do not fully capture the individual user interests and adapt to changing market conditions. Hence, the research developed an adaptive algorithm that uses fuzzy logic, neural networks, and big data to deliver accurate financial recommendations based on user patterns using the Fuzzy Neural Financial Recommendation (FNFinRec) Algorithm. Implemented on a Hadoop platform with a MapReduce framework, the FNFinRec ensures efficient processing of large datasets. Fuzzy K-means clustering is applied, in which fuzzy logic helps to handle uncertainties in financial data and clusters users with similar patterns. An adaptive user profile is developed based on real-time user data input. The Neutral Collaborative Filtering (NCF) recommendation approach aims to predict user interests in financial products/services by learning from user interaction data. The neural networks provide personalized financial recommendations, adapting to changes in user patterns over time for improved accuracy. The metrics such as silhouette coefficient, Davies–Bouldin Index, mean square error (MSE), Precision@k, and Recall@k are used to assess the algorithm’s performance with existing algorithms. The results show that the proposed FNFinRec algorithm outperforms existing methods regarding clustering quality and recommendation accuracy. The competitive processing times that FNFinRec achieves are also crucial for making real-time financial decisions.