Analytics

Data needs to be explored and analysed, and decisions need to be made. These activities are sometimes referred to as descriptive analytics and predictive analytics. This chapter includes details of carrying out comparative tests and cross tabulations and discusses how to detect correlations and patterns in the data that can be useful in selecting variables to build a model. It includes ways to choose the model which has the best quality and is the most stable. When we come to build models, then we can use samples which may be anywhere from 5 000 to 100 000 or more. However, with modern computer power, we can carry out much of our analysis using the whole dataset. Pre‐analytics highlight redundant variables and also likely candidates for removal because of sparseness of different values. Time series readings are likely to be autocorrelated because subsequent values are related to preceding values.