As the technology is evolving day by day, businesses are also evolving and changing their strategies to gain more profit. They are using these developed technologies along with their business experience to get succeed. Especially, in the field of telecommunication where the service providers have to face a large number of customer’s data and this gets more difficult as they cannot focus on each individual customer’s need with respect to services. And if customers are not getting their requirements fulfilled they switch their service provider. So to overcome such problem the idea of churn prediction system comes in picture. Service provider companies are using customer’s data in order to understand and improve Customer Relationship Management (CRM). This paper proposes such various churn prediction systems developed by researchers which uses machine learning approaches that will help telecommunication industries to understand their customer’s need in order to fulfil their requirements. In this paper we proposed churn identification as well as prediction from large scale telecommunication data set using Natural Language Processing (NLP) and machine learning techniques. First system deals with strategic NLP process which contains data preprocessing, data normalization, feature extraction and feature selection respectively. Feature extraction techniques have been proposed like TF-IDF, Stanford NLP and occurrence correlation techniques. Where machine learning classification algorithms are has used to train and test the entire module. Finally experiment analysis shows performance evaluation of proposed system and evaluate with some existing systems.