Having learned about modelling in the previous post, in this post, you will get closely acquainted with CRISP-DM methodology. This is the fifth phase of data mining project, and this is all about evaluation.
CRISP-DM : Evaluation Phase (Step 5)
Evaluating the results
This step deal with the factors like accurateness and overview of the model. In this step, you will be assessing the point in which the model is able to meet your business objectives besides, it will also reveal reasons, for which the model may turn out to be deficient. Apart from this, you can use the real application to analyze the model, considering your budget and time. The evaluation stage also deals with assessing the different data mining results, if any, that you might have generated. These might present additional challenges, hints, and information.
Assessing of the DM results: While evaluating the data mining results, it is necessary to summarize them, considering your business success standard. This should include a final report, concerning whether the task is able to meet the initial objectives of your business.
Approved final models: After you assess the model in terms of business success standard, the models that are generated, and meet the desired criteria, turn into approved ones.
Process of reviewing
The review process is one of the most important stages of the fifth crisp dm evaluation phase. At this stage, the evolving models may appear to satisfy the business needs. Now, you can carry out a more comprehensive review for data mining process. This will help you to determine whether any important task or factor has been overlooked. The review also encompasses the quality guarantee problems. For instance, you need to consider whether the model has been correctly built, whether you have used only the qualities that you were permitted to use and the ones that are accessible for analyses in future.
Reviewing the process: It is necessary to summarize the review and focus on the activities that you have missed out, and the ones you need to repeat.
Determining the subsequent steps
You need to decide the next steps that you are going to take, based on the results of the evaluation and review of the process. You may complete this project and proceed to deployment, set up new projects for data mining or start further interactions. You can simplify the process in the following ways.
List of actions that you might possibly take: You need to create a list of your further actions. Note down the reasons for each, as well as the ones against a particular move.
Decisions: Here, you need to decide on how to carry out the next plans, along with their justification.
Evaluation, the fifth stage of the CRISP-DM process, focuses on the outcome of the assessment.
If you need any sort of support for cross-industry standard process for data mining, you can get across to us, at PGBS. In the next post, you will come to know about the Deployment stage, the last phase of the data mining project.