Pages: 1-6

Abstract

Forecasting is one of the oldest activities and tasks of management and considered as the critical part of every business. Using the experiences and opinions of those who have already purchased a product can be useful in making the right choice. Today, customers use the Internet to buy a product they want. They can find the comments on features, strengths and weaknesses of a product in the Internet. On the other hand, the company and service centers collect comments to forecast the sale of their products. In this regard, the neural networks and data mining models are of useful tools used in developing a high-accuracy model. In present research, a model is presented for forecasting whether the product is purchased by the customer. For this purpose, in the data pre-processing step, the data required are extracted in order to forecast the sales by separating the words and sentences, labeling the components of speech and rooting the words. In the next step, product sales forecasting is performed by using the neural networks and adding 4 effective indicators of advertisements, prices, discounts, free shipping. And finally, the accuracy of 83.65 is reached by boosting the neural networks using the AdaBoost algorithm.