Forecasting is a very important task not only for any business but also in
other fields. There are numerous categories to forecast in business, but determining
the future revenue (revenue forecast) is one of most crucial task so that leader can
propose appropriate policies, decisions to optimize production and business activities.
Revenue forecasting is a complex problem and requires the use of many different
methods and techniques to achieve the highest accuracy. This study mainly analyzes
approaches to select neural network models, processes, highlights the necessary steps
and leverages advancements in deep learning for revenue forecast on a set of revenue
data generated in the monthly, quarterly period from 2013 to June 2019 of 9 regions
in Tra Vinh province. The considered telecommunication services groups including
Internet services, MyTV service, landline phone service and postpaid mobile service
are taken into account for revenue forecast tasks with the deep learning techniques.
The proposed method achieves promising results and is already deployed in the
practical cases at VNPT Tra Vinh.