Classification algorithms require data using category types, while association pattern analysis algorithms require data in the form of attributes whose binary value. Transforming data from continuous and discrete types to binary attributes is called binarization. While transforming data from attribute to attribute category is called discretization. Bineryization and discretization processes are important because good results from this process will affect the performance of data mining algorithms.
The result of binarization can be used for classification, but still leaves a problem if later used for association analysis. Meanwhile, association analysis requires data with asymmetric binary attributes because in association analysis there is only attribute with value 1 that is considered important. Problems with quality category values can be solved by providing each encoding for each category value.
Discretization is usually applied to attributes that will be used in classification work or association analysis. Good depiction also depends on the algorithm used, and this discretization job is also done separately from other data mining jobs.
The result of binarization can be used for classification, but still leaves a problem if later used for association analysis. Meanwhile, association analysis requires data with asymmetric binary attributes because in association analysis there is only attribute with value 1 that is considered important. Problems with quality category values can be solved by providing each encoding for each category value.
Discretization is usually applied to attributes that will be used in classification work or association analysis. Good depiction also depends on the algorithm used, and this discretization job is also done separately from other data mining jobs.
Advertisement
No comments