Nowadays, posting reviews online has become very popular way for people to express their opinions and sentiments toward the products bought or the services received. Analyzing large volume of online reviews available would produce the useful actionable knowledge which could be of economic values to the vendors and the other interested parties. Here, we understood and tried to solve the problem of mining reviews for predicting the product sales performance. The reviews which are posted by consumers involve the sentiments. So, these sentiments expressed in reviews and the quality of the reviews has significant impact on the future sales performance of products or services. And these sentiments are hidden in the Document Corpus which is also known as a Comments Document. The document-level sentiment classification aims to automate task of classifying the textual review, which is given on single topic, as expressing either positive or negative sentiment. Hence by getting these sentiments the overall review or feedback about the particular product can be known in a summarized form which will help vendors to know the overall statistics and the future performance of their product.