In times, when we collect more and more data about individuals, it makes sense to use them for economic research. Publications with the title „Big Data“ are growing since years. This Paper describes the current state of research and shows which steps are necessary to work with unstructured and high dimensional data. Therefore, methods for data structuring will be shown and evaluated. Statistical techniques will be described and used with different datasets. The aim is to show, that simple time-series models are enough to show the usefulness of simple collected datasets for economic indicators. It is also shown, that, if we have unlimited amount of potential regressors, it is possible to get correlations without any causal relation. It is the useful selection on big data what is the most difficult part in that state of research.
ranqingsong / big_data_analysis Goto Github PK
View Code? Open in Web Editor NEWThis project forked from quantlet/big_data_analysis
R Code used in the bachelor thesis "Early Signals for changes of Economic Indicators using Big Data Analysis" (supervised by Lukas Borke)