In previous steps (mostly in Modelling phase) one deals with evaluation of accuracy and generality of generated models, while this step involves evaluation of models with resepect to problem solving objectives. In this phase one has to decide whether there are reasons why built models are deficient. It is also advisable to test models on real problems (test). Besides results directly related to problem solving objectives it is advisable to assess other findings not necessarily related to the original objectives. This might reveal additional information, hints or suggestions for future modelling.
This activity is related to overall assessment of the complete effort in order to find out whether some important issuess might have been overlooked.
This is the step where a decision upon further steps has to be made. If all previous steps were satisfactory and results fulfill problem objectives, the project can move to its conclusive phase, i.e. deployment. If not, another iteration involving data preparation, modelling and evaluation with changed parameters is possible. It is also possible that results of this project suggest a new data mining project.
© 2001 LIS - Rudjer Boskovic Institute
Last modified: September 08 2015 09:28:57.