Mining Frequent Patterns Using Customer Experience

Research Article
Jeet Ganatra, Mihir Thakkar, Kunal Shah and Vaishali Gaikwad
Data mining, Clustering, K-means, MFP, Naïve Bayes Classifier, Sentiment Analysis

Classification and patterns exploration from user data is very crucial for business support and decision making. Timely identification of newly emerging trends is needed in business process. The patterns obtained observing the sales from inventory data point out the market trends and can be used in predicting that has great opportunities in decision making, strategic planning and also in market competition. Some of the algorithms already in use have some limitations. Closed and maximal pattern mining algorithms encounter limitations at mining large patterns, since the process will need to generate an explosive number of smaller frequent patterns. The disadvantage of Naive Bayes classifier when used independently makes a very strong assumption on the shape of the data distribution, i.e. any two characteristics are unrelated given the output class. Hence, the objective of the proposed system discussed in this paper is proposing better methods of decision making for improving sales and services. Identifying the reasons of dead stock, slow-moving and fast-moving products are some of the important mechanisms to be carried out in order to obtain a strong business support, investment and surveillance. Our system proposes an algorithm that can be used for mining patterns of customer data and predicts the factors affecting the sale of products, no matter how huge the customer data is. This system combines two algorithms that are K-mean and Most Frequent Pattern (MFP) algorithm. The experimental results showed that the proposed hybrid k-mean plus MFP algorithm can generate more useful patterns from the large stock of data. The final step is to provide a sentiment analysis by using the Naive Bayes Classifier based on customer reviews and feedback.