Defect prediction relies on data and modelling techniques. Previous work has predominantly been focused to improving the latter. Despite numerous studies exploiting a wide range of machine learning techniques, prediction accuracies improve at a slow pace. On the other hand, the focus on data has somewhat been limited. High-quality data is difficult to obtain and mostly consists of metrics which have been lingering round for a long time. Due to the influx of digital data, other fields have developed some interesting techniques for its analysis. In this journal club I will discuss some of those techniques for data analysis and link them to their possible use in defect prediction.