Numeric Missing Value´s Hot Deck Imputation Based on Cloud Model and Association Rules

Filling missing value is main task of data-processing, at present Hot Deck Imputation is preferred. Defining the similar standard of Hot Deck Imputation objectively becomes an important prerequisite. The Cloud model combines ambiguity and randomness organically to fit the real world data objectively. first get the cloud models which present the raw no missing value, then to discrete the numeric value and do the association rules mining in the discrete value to get the knowledge base, filling the missing value with the value which generated by the cloud model from the knowledge base. The method considered the original data´s distribution as a whole and to improve its precision with association rules from the raw data for each record, it simulates the humans´ behavior; this method has smaller absolute mean difference than other methods.