An Exploratory-Descriptive-Inferential Analytics Approach to Characterize Failure-Causing Problems in a Single and Multiple Police Functions

Abstract

Data analytics is fundamental in policing since it provides insights to different policing issues. One of these issues is failure-causing problems in police functions. Failure in any police organization is a problem. An exploratory-descriptive-inferential analytics approach is proposed to help police senior management to look deeper into a failure and find out why it happened, how often, and to predict which failure will most likely occur in the future. It also provides them with the information needed to support the decisions necessary to reduce and prevent failure occurring in police functions. There are many different causes for different kinds of failures. There are a variety of different types of analytical methods currently in use in organizations and in settings outside policing. Exploratory-descriptive-inferential analytics approach is a combination of exploratory, descriptive, and inferential statistical techniques related in a whole. It involves collecting and analyzing complex policing datasets, and using analytical methods to turn them into actionable information and insights related to the issue causing the failure. The approach uses Exploratory Hierarchical Cluster Analysis (EHCA) in conjunction with Measure of Central Tendency (MCT) and Linear Regression Modeling (LRM). These methods are simple, objective, and unquestionable with no personal interpretations or judgments are involved in the analytical procedure: data objects are analyzed according to statistical properties. Due to a lack of access to a real failure frequency data set, the different stages of the approach are explained on a fictitious generated dataset with the intention of transfer learning to real data. The data set consists of 12 typical police functions and 45 categories of failures that could possibly occur during typical day to day duties. This enables us to explain how the exploratory-descriptive-inferential analytics approach should be used in a valid and understandable manner as simply as possible and to create realistic methodological procedure for users/readers. The results show that the suggested approach is able to group police functions into patterns based on categories of failures found in them, able to identify recurring categories that existed in each police function with plot summaries about the mean values, and also able to predict the values of what failure is likely to occur in the future in a given police function. This gives us confidence that if the current approach is applied on a real complex data it will give empirically-based, objective, and replicable results which can be used to identify the root cause of a failure and why it happened, and to indicate whether or not certain failure categories are repeatable in one police function or another.