The Deep Feature Synthesis (DFS) algorithm automates feature engineering and is capable of extracting and applying complicated featuresto a variety of processes. Due to the novelty of DFS as a method for feature engineering, critical ways for dealing with missing values and unwanted data in a dataset have yet to be established. This paper discusses the usage of mean, median, and mode imputation to preprocess data before analyzing it.However, it is only limited to displaying the differences between nonimputed and imputed datasets. This strategy enables users to obtain more precise results by eliminating biased estimations. This study demonstrates that there is a distinct difference between the two datasets. This paper is concluded by proving that imputing datasets will cause distinctness in the results compared to the results of the datasets with missing and unwanted values.

Keywords : Deep Feature Synthesis, Auto Feature Engineering, Imputation