Data use becomes equally essential in the current age of big data, as an enormous quantity of data is produced every second. Noise is a blatant reality in real-time data. Due to the abundance of data discrepancies and the variety of noise sources, training machine learning models in noisy environments can be considered an arduous task. Distinctively, it has a significant impact on the domain of data analytics, as data cleaning is a very large part of the data processing cycle. Noisy data leads to inaccurate training of models and produces wrong results, thus complicating the analysis of the concerned data. Furthermore, the presence of noise significantly impacts the prediction of meaningful data, decreasing classification accuracy and producing poor results. The problem of noise has attracted great attention considering our ongoing transition to a world where datasets exponentially gain importance. This paper gives us an understanding of “data imputation,” a method used for handling missing values in a dataset. We focus on the types of imputations and how different classifiers perform in them. For analysis, we have collected a public data set, namely the Melbourne housing dataset from Kaggle, for our research. We have also used Iris Dataset from Scikit-learn.