Oryza Wisesa - Department of Electrical Engineerng, Universitas Mercu Buana, Jakarta, Indonesia.
Andi Andriansyah - Department of Electrical Engineerng, Universitas Mercu Buana, Jakarta, Indonesia.
Osamah Khalaf - College of Information Engineering, Al-Nahrain University, Baghdad, Iraq.

Sales prediction analysis requires intelligent data mining techniques with accurate prediction models and high reliability. In most cases, business highly relies on information as well as demand forecast of the sales trends. This research uses B۲B sales data for analysis. The B۲B data could provide information on how telecommunication company should manage its sales team, products, and budgeting flows. The accurate estimates enable Telecommunication company to survive the market war and increase with market growth. Comprehensible predictive models were studied and analyzed using a technique of machine learning to improve the prediction of the future sale. It is hard to cope with big data and sale prediction accuracy if the system of traditional forecast is used. In this study, machine learning technique was also used to analyze the reliability of B۲B sales. In addition, at the end of this research, other measures and techniques used to predict sales were introduced. The predictive model with best performance evaluation is recommended to forecast the trending B۲B sales. The study results are put into an order of reliability and accuracy of the best method to predict and forecast including estimation, evaluation, and transformation. The best performance model found was Gradient Boost Algorithm. The result form graph the data close together from beginning till end of data target MSE and MAPE result are the best result than other method, MSE =۲۴.۷۴۳.۰۰۰.۰۰۰,۰۰ and MAPE =۰,۱۸. This model performed maximum accuracy in predicting and forecasting of the future B۲B sales.

B۲B, Business to business, Economy Engineering, Machine Learning Techniques, Sales Forecasting, Prediction, Telecommunication, reliability