Hybrid model for data imputation: Using fuzzy c means and multi layer perceptron

Database store datasets that are not always complete. They contain missing fields inside some records, that may occur due to human or system error involved in a data collection task. Data imputation is the process of filling in the missing value to generate complete records. Complete databases can be analyzed more accurately in comparison to incomplete databases. This paper proposes a 2-stage hybrid model for filling in the missing values using fuzzy c-means clustering and multilayer perceptron (MLP) working in sequence and compares it with k -means imputation and fuzzy c -means (FCM) imputation. The accuracy of the model is checked using Mean Absolute Percentage Error (MAPE). The MAPE value obtained shows that the proposed model is more accurate in filling multiple values in a record compared to stage 1 alone.