International Research Journal of Engineering and Technology (IRJET)
e-ISSN: 2395-0056
Volume: 09 Issue: 11 | Nov 2022
p-ISSN: 2395-0072
www.irjet.net
CUSTOMER CHURN PREDICTION Snegha K*1, Karthika M I *2 ,Deepika Dharshini S*3, Janani B S*4 *1Student,
Department of Information Science and Engineering, Bannari Amman Institute of technology, Erode, Tamilnadu, India *2Student, Department of Information Science and Engineering, Bannari Amman Institute of technology, Erode, Tamilnadu, India *3Student, Department of Information Science and Engineering, Bannari Amman Institute of technology, Erode, Tamilnadu, India *4Student, Department of Information Science and Engineering, Bannari Amman Institute of technology, Erode, Tamilnadu, India ---------------------------------------------------------------------***--------------------------------------------------------------------human behavior. AI systems unit of measurement wish to Abstract - Churn rate is something used to predict the
perform difficult tasks in an exceedingly} very technique that is like but humans solve problems. Deep learning might be a collection of machine learning, that's primarily a neural network with three or heaps of layers. These neural networks conceive to simulate the behavior of the human brain albeit removed from matching its ability permitting it to “learn” from large amounts of data.
number of customers leaving a particular company. It is used to maintain a sustainable customer-company. Deep learning method is used to analyze the churn rate and process huge amounts of customer data. In this paper, a deep learning method is used to predict the number of customers who will be retained in the industry and will be churned out. The model used here is Artificial Neural network model, this model is the most used in all the churn rate prediction. Machine learning is a type of deep learning model that uses neural networks multilayer architecture. The artificial neural network is based on the collection nodes we will call the artificial neurons, which further model the neurons in a biological brain. The results of the models were compared with accuracy classification tools, which are precision, recall etc. The results showed that the deep learning model achieved better classification and prediction success than other compared models. After prediction of the result the final data is displayed in the dashboard for live visualization. This paper mainly focuses on the churn rate prediction using machine learning and deep learning models and analyzing the final result using a live dashboard display using python libraries.
1.1 LITERATURE SURVEY
Key Words:
Churn rate, Deep Learning, Machine Learning, Artificial Neural network, Prediction.
Ning metallic element [1] projected the employment of boosting formulas to spice up a consumer churn prediction model at intervals that customers unit separated into two clusters supported the burden assigned by the boosting algorithmic program. As a result, a high risk consumer cluster has been found. provision regression is used as a basis learner, and a churn prediction model is made on each cluster, severally. The experimental results showed that the boosting formula provides a wise separation of churn information compared with one provision regression model.
1. INTRODUCTION Churn prediction implies that detection that customers unit of measurement probably to depart a service or to cancel a subscription to a service. Churn studies are used for years to understand chance and to see a property customer-company relationship. Deep learning is one in each of the fashionable methods used in churn analysis as a result of its ability to technique large amounts of consumer info. throughout this study, a deep learning model is planned to predict whether or not customers at intervals the retail business will churn at intervals the longer term. It is a crucial prediction for many businesses as a results of accomplishment new shoppers sometimes costs over holding existing ones. Machine learning might be a subfield of AI, that's usually printed as a result of the potential of a machine to imitate intelligent
© 2022, IRJET
|
Impact Factor value: 7.529
P.C.Pendharkar [2] taught two Genetic Algorithm(GA) based totally neural network (NN) models to predict the consumer churn. the first GA-based NN model used a cross entropy based totally criteria to predict consumer churn, and additionally the second GA based totally NN model created some efforts to directly increase the prediction accuracy of consumer churn. exploitation real-world consumer dataset and three various sizes of NNs, they compared the two GA-
|
ISO 9001:2008 Certified Journal
|
Page 154