Towards data self-awareness: Automated aggregation, geometrization, predictability and cross-validation of real-world data

We propose a strategy for automated assessment of self-awareness of real-world data sets based on automated aggregation, geometrization, predictability and crossvalidation. The strategy was tested using three different scenarios: urban spatial structures acquired from the OpenStreetMap in which We propose computationally feasible statistical algorithms for the automated assessment of isolation and integration of urban locations and neighborhoods by using maps acquired from the Open- StreeMap service; Standard & Poor’s 500 financial time series in which we showed that inhomogeneous density of states in a discrete model of Standard & Poor’s 500 phase space leads to inequitable predictability of market events, most frequent events might be efficiently predicted in the long run as expected from Mean reversion theory; and Maintenance and Flight data provided by the Department of Defense.