Load the auto-mpg sample dataset from the UCI Machine Learning Repository (auto-mpg.data) into Python using a Pandas dataframe. The horsepower feature has a few missing values with a ? - replace these with a NaN from NumPy, and calculate summary statistics for each numerical column (Hint: Use an Imputer from Scikit). Replace the missing values with the overall mean, median, and mode (Hint: Pandas makes this easy) - and calculate the variance of the feature. What imputation results in the lowest variance? Why? Is there a different method of imputing values that would match the distribution more accurately? Describe your method. Load the auto-mpg sample dataset from the UCI Machine Learning Repository (auto-mpg.data) into Python using a Pandas dataframe. The horsepower feature has a few missing values with a ? - replace these with a NaN from NumPy, and calculate summary statistics for each numerical column (Hint: Use an Imputer from Scikit). Replace the missing values with the overall mean, median, and mode (Hint: Pandas makes this easy) - and calculate the variance of the feature. What imputation results in the lowest variance? Why? Is there a different method of imputing values that would match the distribution more accurately? Describe your method