Multivariate time series prediction of cellular traffic via Machine Learning

We describe all the steps needed to build an original real-world dataset used to perform the so-called Multivariate time series prediction through Machine Learning techniques. The dataset contains several features of cellular traffic organized into time series. The goal is to exploit Machine Learning to predict the future behavior of a feature based on the past value of the feature itself and on the past values of other features.