Using Gaussian Process to Predict Five Personality Traits and Facets Using Data Collected from Smartphones

With globalization and the 4th industrial revolution, the world has become emerged in technology. A smartphone is now capable of holding an entire identity of a person within. It is easy to interpret human behavior using the behavioral data collected from a smartphone; therefore, it presents the opportunity to predict personality. This study focuses on predicting the big five personality traits and their thirty facets. Using multi-class Gaussian Processes, we present both effects of regression and classification through the Gaussian process on the high dimensional data collected from smartphones. The use of the Gaussian process in machine learning is underappreciated. Our results indicate that Gaussian Process can provide better confidence in the model by calculating the uncertainty while providing promising results in agreement with the previous literature. Our cross-validated results are then compared with the results provided in the previous literature, results obtained at random, and a baseline model. We also compare these results with other baseline models that show potential for promising results or those that have been previously modeled. This study also focuses on the benefits including mental wellness as well as the dangers like psychological targeting and privacy concerns, aroused by the collection and modeling of behavioral data acquired from smartphones.