The Europe 2020 strategy prioritises Smart Growth and proposes to develop an economy based upon knowledge and innovation. This initiative has identified the insufficient use of information and communication technologies as an inhibitor to Europe's growth. The 2013 Work Programe for Information and Communication Technologies targets the development of methodologies for analysing extremely large volumes of data, commonly known as big data.
Big data does not mean always mean useful or complete data. The definition of data mining is the extraction of hidden useful information within data. It is a powerful technology with the potential to help businesses and organisations make effective decisions and operate more efficiently.
Data mining can be applied to big and small data. In cases where big data is highly aggregated it can be applied to the small, incomplete data sets to reduce the big data computation. Therefore methods of data mining which can compute small, incomplete data sets benefits not only macroeconomics, climate change evaluation and R&D of new products but also big data analysis.
In comparison to big data analysis, little research has been done into data mining of small data. Specialised work into speaker recognition and bioinformatics has been conducted although these are not translated to other areas.
The emerging subect of Grey Systems has become a very effective theory for solving problems within environments with a high degree of uncertainty containing small, discrete, incomplete data sets. The theory pays particular attention to gaining maximum use from limited data and has the capcability to establish reasonable forecasting models using only a few recent data samples.
The theory of Grey Systems has numerous successful applications; energy, hydrology science, communications, environment, software, business, reliability, manufacturing, water resource, economy, stock control, medicine and R&D management. As an emerging technology Grey Systems have achieved significant progress in China, and its applications have been seen in nearly each section of industry and society. However research in this area is just beginning in Europe and has limited applications outside of China and East Asia.
Limitations in grey forecasting and decision models have been identified. In grey forecasting there are many available models, each one has different merits and feasibilites under different conditions. In tandem the calibration of any selected model has a significant impact on the results. However the selection and calibration of a model to an application requires expert knowledge. Exisiting grey forecasting models are also limited to integer-order sequences although fractional-order sequences are possible in the real world.
In current grey decision making models, the reward is assumed as a known (white) number, where in reality this is more likely to be an uncertain (grey) number. Existing decision models do no take into account the relative significance of information from different sources or the impact of past and future rewards.