What is an appropriate way to handle missing values in a column when the missing values exceed 50% (of a descriptive feature) and the dataset is limited when building a machine-learning model?

a. Remove the column entirely from the dataset.

b. Impute the missing values using the mean or median of the column

c. Impute the missing values using the mode of the column

d. Impute the missing values using a model