Question

Comment critically: “Data are an important component of an empirical analysis. It should be complete and clean. Data issues may derive the choice of the model and even lead to the alteration of the theoretical model.”

Answer 1

• A complete and clean data is what is ideally required. It helps in developing the best possible model.
• However, even if there are some missing values, we can still perform the regular algorithms on the data once the missing values are imputed.
• If only a small % of the data, say 5% or less is missing, then it is acceptable to drop those observations and proceed with the analysis with the remaining complete observations.
• If, however, a larger proportion is missing, it becomes necessary to impute the data. Data imputation can be done by various methods like mean imputation, median imputation, regression imputation, multiple imputation imputation via chained equations, etc. depending upon the type of variables.
• When imputation is done appropriately, the prediction/ classification results obtained from the model are reliable.

Hope this was helpful. Please leave back any comment.