Exam DP-100 topic 2 question 38 discussion

The MICE (Multiple Imputation by Chained Equations) method is effective for imputing missing values by considering the relationships between different columns, making it suitable for datasets with many missing values

CoPilot game me this ans: D. Replace using MICE (Multiple Imputation by Chained Equations) MICE is an effective imputation technique that can handle multiple columns with missing values by using regression models to iteratively impute the missing data, making it suitable for your needs.

The point is "The data does not require the application of predictors for each column." So, it means reducing the dimension and use the PCA - Primary Component Analysis

Question is outdated in my opinion. In Clean Missing Data, you can see only: Replace with meanmedian/mode/ and remove enitre row/column https://learn.microsoft.com/en-us/azure/machine-learning/component-reference/clean-missing-data?view=azureml-api-2

A is the answer. It is in classic version.

As a predictor is not required we can use PPCA method

Answer is PPCA. MICE is wrong totally check documentation.

D. Replace using MICE (Multiple Imputation by Chained Equations) is a method that should be used to clean missing data in this scenario. It is commonly used when the data has missing values and the aim is to impute the missing values while preserving the relationships among variables in the data. A. Replace using Probabilistic PCA (Principal Component Analysis) is not the most suitable method for cleaning missing data in this scenario, as it is typically used for dimensionality reduction and feature extraction, rather than imputing missing values. The method of choice for cleaning missing data in this scenario is D. Replace using MICE (Multiple Imputation by Chained Equations), as it is commonly used for imputing missing values while preserving the relationships among variables in the data.

PCA is for dimentionality reduction: it diagonalize the covariance matrix (being simmetric for the spectral theorem u can always diagonalize it) and than cuts off the dimensions with small eigenvalues/variance... I am not aware of any variant of this algoritm to impute missing values. do you have any reference? thank you

MICE vs PPCA, this is not so easy to answer in practice, for exam purpose, I agree with A

A, is the correct answer. For dimension reduction. PCA algo is significantly used.

What does "application of predictors" for each column mean?

Yes, given Answer is right

Agree with the answer