An Efficient Approach for DW Design and DM in Crime Data Set

Data Warehouse (DW) represents the repository of data on which Data Mining (DM) techniques are applied to discover valuable knowledge. DM represents a wide range of tasks and techniques that represent the core of what is known as Knowledge Discovery in Database (KDD). Crime Analysis is an important application of DM, where data from different applications and sources are analyzed to extract and predict knowledge concerning crimes and criminals aiming to prevent and avoid crime occurrences. This paper presents a solution for DW design in three different models (Star, Snow flake and Galaxy) and a model for classifying the data sets related to crimes, offences and criminals aiming to predict some knowledge explaining the crimes trends, criminal groups, and related features. The result from this paper tends to help specialists in discovering patterns and trends, making forecasts, finding relationships and possible explanations, mapping criminal networks and identifying possible suspects. Different DW models were suggested since each model has its own advantages in data analysis by providing better mining algorithms performance. Data Mining techniques are used to analyze the logged data. One of the most common and effective DM technique is Classification. The classification is based mainly on grouping the crimes according to the type, location, time and other attributes, and grouping criminals according to their age, job, income, education, history and other attributes. Using different DW models showed efficient analysis process in both normalized and data reduction disciplines in both Snowflake and Galaxy DW models. Free data available on the Internet from some police departments are the source of the data about the crimes and the criminals and they were used to create and test the proposed framework, and then these data were preprocessed to get clean and accurate data using different preprocessing techniques (cleaning, missing values and removing inconsistency). The preprocessed data were stored in three different DW models to find out different crime and criminal classes, groups, and clusters. WEKA mining software and Microsoft Excel were used to analyze the given data. Decision Tree and Rule Base Algorithms were used for classifying and predicting the crimes, criminals and offences groups.