A Machine Learning Specialist is developing a regression model to predict ticket sales for an upcoming concert. The historical ticket sales data consists of more than 1,000 records containing 20…

Dropping a small number of unusual, but plausible, records (33 out of 1000) can prevent them from disproportionately influencing a regression model sensitive to outliers. Replacing blank numerical values with the mean is a standard and effective imputation technique that preserves data points while providing a reasonable estimate for missing information.

Normalizing unusual data scales the values but does not inherently reduce the leverage or influence of outliers on a regression model. While creating a Boolean variable for blank values is a valid technique, it typically complements, rather than replaces, a primary imputation strategy.

While dropping unusual records might be acceptable, filling blank numerical values with 0 is often inappropriate unless 0 has a specific, meaningful interpretation for that variable, as it can otherwise introduce significant bias.

Using unusual data as-is without modification can negatively impact a regression model's performance due to its sensitivity to outliers. Similar to B, creating a Boolean variable for blanks is helpful but doesn't fully resolve the missing data problem without an imputation strategy for the original column.