Accuracy Assessment of forecasting services
Tutor / director / evaluatorFranch Gutiérrez, Javier
Document typeMaster thesis
Rights accessOpen Access
English: A service system is a dynamic configuration of people, technologies, organisations and shared information that create and deliver value to customers and other stakeholders . The following cases are examples of customers receiving a service: taking a bus to go somewhere, or going to a restaurant to have a meal, or for a small IT (information technology) company, contracting a service to a bigger one in order to save costs and time. Service-oriented architecture (SOA) has become more popular during last years. Basically, this emerging development paradigm allows service providers to offer loosely coupled services. These services are normally only owned by the providers. As a result, the service user or client does not have to worry about the development, maintenance, infrastructure, or any other issue of how the service is working. To sum up, the user just has to find and choose the proper service. On the one hand, it presents several advantages. Firstly, common functionality can be contracted as a service in order to be able to focus on the own core missions. Secondly, it decreases the cost, since it is cheaper to contract a service than creating it yourself. Thirdly, clients take benefit of provider’s latest technologies. On the other hand, there is one big drawback: lack of trust. When you contract a service, you lose the direct control, the provider has access to your own data, you depend on him, and you experiment delays since your functionality is not working in-home. That is why the user has to decide previously which service is the most appropriate for his needs. Each client has different needs: quality (it varies among services), reputation (a famous or recommended provider usually gives more confidence), speed (agreements not to break thresholds), security (contract and trust in the provider), personalisation (preferential treatment from the provider), and locality (law is not the same in all countries). Therefore, a customer needs to know about the best service(s).Among all kind of services, we concentrate on forecasting services. Forecasting services show in advance a condition or occurrence about the future. There are plenty of domains: weather forecasts, stock market prices, results in betting shops, elections… Let us see a domain which is really familiar to all of us: weather forecast. When we are planning to travel, going somewhere or just deciding what to wear first thing in the morning, we wonder about weather conditions. To make these decisions, we check the weather forecast on TV news, a thermometer, or on a web site. However, sometimes we check several predictions and they do not agree. Which one will be the most accurate? Our goal in this master thesis is to assess the accuracy of these forecasting services in order to help prospective users to choose the best one according to their needs. To do it, we are going to compare forecast predictions with actual real observations.