Big data in physics, biology and social science

In the digital age, we are creating 2.5 billion gigabytes of data every single day of our lives, and the 75% is unstructured, coming from texts, videos and opinions spread through online social networks. These data sets are so voluminous and complex that traditional data analytics methods are often inadequate to extract value from them.
Here, I show how to extract summarized data by identifying an optimal set of network nodes, called influencers, with high predictive power in anticipating opinion trends and tracking the effectiveness of messages in massive social networks. I illustrate the theoretical framework to identify influencers using the concept of optimal percolation (OP). Big data analyses in social networks reveal that the set of influencers identified by OP theory is much smaller than the one predicted by heuristic centralities. Remarkably, a large number of previously neglected weakly connected nodes emerges among the optimal influencers.
I show that OP theory is also a powerful indicator of personal economic status. Said in very simple terms, we discovered that people located in positions of high influence have high financial levels. This result comes from the combined analysis of two large-scale datasets including telecommunication and financial data of 110 million people in Mexico.
Finally, I also discuss applications of OP theory to brain networks. In particular, I show how OP theory identifies the essential nodes for integration of the memory network in a rat brain.