Feature weighting as a tool for clustered based imputation model

Imputation of missing attribute-values is an important data pre-processing step. Missing of the attribute values are an inevitable problem in real-world data collection. However, most of the data processing models don’t have any mechanism to deal with missing values. One of the solutions to this problem is to have an imputation step added in data pre-processing. Imputation model relays on the initial prediction of the missing values and has no mechanism to distinguish the observed and imputed values. But, the imputed values are only as good as the assumption used to create them. In this research, we introduce an unsupervised cluster-dependent feature weighing imputation model. This model uses the feature weighing factor to rescaled the data to nullify the effect of initial prediction of the missing attribute-values.

Date: 17/02/2017
Time: 16:00
Location: LB252

Share this post on: Twitter| Facebook| Google+| Email