A Hybrid Modified Deep Learning Data Imputation Method for Numeric Datasets

Missing data is a major problem in terms of both machine learning and mining methods. Like most these methods do not work with missing data, negative results may occur on the performance working ones, also. Imputation preprocessing method used to replace appropriate values. This study aims at developing hybrid modified imputation based deep approach. For this purpose, we use Random Forest Datawig (called RF-DLI) together.  learning-based library that supports value for all types data. RF-DLI approach includes following steps impute First, importance each attribute dataset determined (RF). Second, important 50% attributes are selected. Finally, imputed datawig (DLI) using attributes. The uses six real-world datasets from different fields 30% compared KNN, MICE, MEAN approaches MAE, RMSE, R2 evaluation metrics. show cases, has better than other techniques mentioned.