In this video, I explain how to use np.where in data preprocessing. One way to utilize np.where is to replace zeros in data with np.nan, which allows one to use pd.fillna to fill the values with the mean/median/mode of the column. Although you could also do this in Pandas with pd.replace.