An Algorithm for Selecting a Data Mining Technique

Selecting a data mining technique is an important step in the data mining process. Techniques like association, clustering, regression, naïve Bayes, and time series may be used for data mining. However, the various tools available in the market do not award the user a chance to verify whether the tool is appropriate for their data. This study presents an algorithm that may be used to select a technique based on the structure of the data to be mined. Literature was reviewed to identify the factors that may be considered in selecting a technique. The spiral model was adopted for development of the algorithm. The algorithm compares the data source to the defined criteria which has weights assigned to determine the suitable technique. A score is allocated to each evaluated technique and the technique with the highest score is recommended. The scoring and weighting details are described in pseudocode and flowcharts while Java programming language was used to implement the algorithm. The resultant artefact suggests a data mining technique after analysing the structure of given a data set.