Necessity of customer inputs for online group shopping using Support Vector Machines

Online group shopping is feasible if the online store comprises useful products. The product needs to attract customer interest. The customer will have varied choices to choose from the available products. For feasibility of Online group shopping by customers the products have to be properly classified. The site offers the customer to form a group based on his/her interest. The site gives the best deal on a product, the threshold or the limit to form a group is proposed by the site. The basic problem with Online stores is that they offer discounts to customers without knowing the customer interest on a particular product. The gap of Online group shopping is rectified by taking feedback from the customer when the customer visits the site. Discounts are offered to the customer based on customer interest. In this paper the focus is on group of people interested in a particular product from among the ones displayed on the site and predict the potential buyer based on feedback taken. The paper uses Support Vector Machine for generating results. Paper further compares SVM with C4.5 algorithm. The comparative analysis of true positive and true negative is 98% and 98% respectively for SVM after testing, C4.5 giving 76% for true positive and 98% for true negative. It's known that SVM works well for linear classification. SVM is not about “yes/no” classification, but it is trained to give numerical values. The paper deals with proper classification of data or products. The paper further proves that good classification leads to better retrieval of information. The question is, are the customers able to specify the product being purchased? If yes a feedback is taken. Second question being once the customers specify the product is the product information specified with fast response. The number of feedbacks are used for generating a threshold. Based on threshold values discounts are offered to customers leading to better business. The paper shows SVMs performs better than C4.5 even in the presence of some noise.