Exam Certified Machine Learning Professional topic 1 question 2 discussion

It's two-way b/c two-way doesn't rely on counts and, as long as the nulls are not evenly distributed over all categories (they are only in a single category in this scenario) it will catch this difference. A one-way is dependent on counts, so if there are nulls in one sample, it will see that as a different distribution when in fact it may not be (distribution could be the same as before, only with less counts b/c of nulls).

It seems like we want to determine if there's a statistical difference in one direction. Ex: between last month's data and new data, particularly observing more blanks for a specific categorical variable. It seems like a one-way Chi-squared Test is the most appropriate, what do you say?

As the question is about one input variable, so my answer would be B.

It seems like we want to determine if there's a statistical difference in one direction. Ex: between last month's data and new data, particularly observing more blanks for a specific categorical variable. It seems like a one-way Chi-squared Test is the most appropriate, what do you say?

C. Two-way Chi-squared Test This tool can help the engineer determine if there is a statistically significant association between the time period and the presence of missing values in the categorical variable.

Would the answer not be C? One-Way Chi-Squared Tests will pick up changes in the overall count of null values as a shift in distribution, even if the nulls are distributed evenly among the category values. The question specifies that they are looking for a shift in a particular class value distribution, which would be better for Two-Way Chi-Squared Test

Since it's just one categorical input variable over time, the one-way Chi-squared test would be the more appropriate choice.

The correct answer is B. One-way Chi-squared Test. The Chi-squared test is a statistical hypothesis test that is used to determine whether there is a significant association between two categorical variables. In this case, the two variables could be the presence (or absence) of a value and the time period (old data vs. new data).