KDD Process

• Data selection - selection criteria is applied to the data to create subsets of data that is needed.
• Data cleaning and preprocessing - removal of any incorrect data. Joining data into a more amenable data mining table.
• Data transformation - e.g. converting data types, grouping data, defining new attributes.
• Data mining - searching for patterns via formulated hypotheses.
• Interpretation and evaluation - patterns interpreted into knowledge for decision making. May require that data mining is repeated with other methods.

Data held in a data warehouse has already been through the mill of cleaning and transformation. Therefore the data may not need to go through the data cleaning and preprocessing or data transformation stages.

Comments, suggestions, ideas to
Stuart Banner