Forecast, as an activity of human cognition, has existed since ancient times. Specifically, it refers to scientific speculation on possible trends and levels of things in the future by using various qualitative and quantitative analysis methods according to the objective process and laws of the past development and changes of things, and the current movement and changes of things. The forecast falls into three types: economic forecast, technical forecast and demand forecast. In time dimension, it can be divided into short-term forecast, medium-term forecast and long-term forecast. From the perspective of forecasting method, it can be divided into qualitative forecast and quantitative forecast. Qualitative forecast belongs to subjective judgment that is based on estimation and evaluation. Quantitative forecast is a method that can be used to predict the future in historical data related to the past on the basis of setting. Historical data may contain factors such as trends, seasons, and cycles.
The advent of the 5G era marks the beginning of a new digital era. The accelerated development of science and technology has brought an endless stream of new things, and there are more uncertainties in the future than in the past. Therefore, it is more important to make a good rolling plan of the business. Data analysis and business forecast are the basis of rolling planning. According to historical data and experience, future business development can be predicted and a quantitative analysis can be provided. The new intelligent big data platform facilitates fast learning iterations in massive data and timely correcting forecast data to improve accuracy. This provides effective data support for operators' strategic development.
The global large-scale 5G deployment is planned for 2020, but the outbreak of Covid-19 has caused great difficulties to global social and economic development and disrupted the existing rhythm for the telecom industry. According to authorities, the number of wireless users worldwide will reach 8.72 billion by 2025, a decrease of 99 million compared with previous forecasts. From 2019 to 2025, the investment curve of 5G will be smoother than that of 4G, and the investment schedule will also be delayed. The first reason is that the revenue growth of 5G is slow, with a compound growth rate of 1.3% during this period, which is lower than the 2% compound growth rate of 4G revenue in the period from 2011 to 2019. Secondly, the high cost of 5G network infrastructure also restricts operators' enthusiasm for investment.
The above is a macro analysis of the telecom field. In specific areas and markets, how can we make a good forecast to provide strategic support for business development of operators.
Operators accumulate massive data in their daily operation, including operation data, business data, network data, user data and other basic data. In the 5G era, new services emerge one after another, and there are more uncertainties in the future.
In these uncertainties, how can operators forecast and plan future business, develop valuable users, do a good job in daily operation management, and improve their return of investment? It is known that the core value of big data lies in forecast and the core of enterprise operation is to make correct judgments based on the forecast. Big data forecast is based on big data and the forecast model to forecast the probability of something in the future. The biggest difference between big data and traditional data analysis is to shift the analysis from being oriented to the past that has happened to the future that is about to happen. Based on these massive big data, using new artificial intelligence technologies, operators connect existing business domains, management domains and operation domains to build a unified intelligent digital platform for value-centric operation (Fig. 1). With the intelligent big data platform, they can improve their operation level and provide timely and accurate forecast capability for future business development. The gold value of big data is thus highlighted.
Intelligent big data platform can provide tightly coupled AI capabilities such as data analysis, scene recognition, model design training, model/algorithm library, reuse annotation management, and monitoring services. Through the business sharing, the service platform builds a core business center, shares service capabilities of business units, and provides rapid business combination. It achieves accurate forecast of business, capacity, coverage, and revenue, providing strong support for precision marketing of operators.
In practical applications, based on business and user data analysis, accurate user insight is first achieved by abstracting the tagged user model from information such as user social attributes, living habits and consumption behavior. Upon the user insight, users are then classified. Through the user insight and classification, different types of users can be clearly and accurately located. The training model is used to make accurate user forecast, recommend appropriate products, track customer feedback, and complete business closed-loop optimization.
There is a use case to see how to use the intelligent big data platform to complete business analysis and recommendation.
Step 1: User insight.
Know the details of users' business usage by analyzing the business type, proportion, time distribution, and geographical location distribution.
Step 2: User analysis.
Classify different types of users and find out their consumption habits and preferences through user insight.
Step 3: User forecast.
Bring classified users into the model for iterative analysis according to their consumption habits and preferences, analyze and forecast emerging businesses and tariff packages that users may be interested in, and provide business reports that best match users' propensity to consume.
Step 4: Business recommendation and feedback.
Make business recommendations according to different types of users, give feedback by regularly tracking users' new business usage, carry out iteration and constantly modify the model, and form a closed loop.
In this case, through user insight and accurate forecast, business sales conversion rate, actual revenue and user satisfaction are all improved to achieve a win-win situation for both users and operators.
Under the guidance of users and business planning, current network operation data can be analyzed to achieve network insight. Accurate network planning is made by accurately forecasting network capacity, coverage, performance and traffic. In network operation, information about network performance and customer experience can be tracked in real time to complete a closed-loop optimization of network planning. This enables operators to build networks quickly and accurately, seize the golden window period, and deliver accurate services for value areas and customers, thus gaining an advantage in the competition.
There is another use case to see how to use the intelligent big data platform to make accurate 5G network planning.
Collect data such as user distribution, traffic distribution, business type distribution, as well as user experience and complaints based on existing 4G network data to form a network operation dashboard.
Bring data into the existing model for business, capacity and coverage analysis and forecast as well as user experience analysis and evaluation based on the network operation dashboard.
Forecast 5G network service, capacity and coverage, plan hotspot areas and select wireless sites based on the above analysis and evaluation.
Carry out data simulation to find out over-coverage and weak coverage areas, and adjust wireless sites (remove redundant sites and add weak-coverage sites) based on the selected sites and forecast data, and also evaluate the site planning base on the results of forecast and simulation.
Generate site planning solutions and evaluation reports automatically to complete accurate 5G planning based on the above analysis and evaluation.
The above are the solutions to two scenarios such as business recommendation and 5G network planning using an intelligent big data platform. In practical applications, the intelligent big data platform can play an important role in user forecast, business forecast, revenue forecast, market size forecast and so on.