Microsoft Certified: Azure Data Scientist Associate - (DP-100) Exam Questions

A- Detection of data drift in datasets

B- A machine learning model's performance metric drops below a specified threshold

C- Remaining impurities in data after completing preprocessing and wrangling

D- Registration of a new machine learning model in the workspace

A- Feature selection

B- All options listed

C- Data normalization and transformation

D- Handling missing values

A- In the Azure Machine Learning workspace storage, alongside other assets like notebooks and logs

B- Co-located with the model code in the repository

C- In the code repository under a directory named .github/workflows

D- Within Azure DevOps storage, specifically for pipeline definitions

A- Outputs trained data

B- Provides access to the datastore

C- Passes prepared data to the training step

D- It is redundant and should be part of the data prep step

A- Adequate unused capacity is essential to handle the training jobs without impacting other operations running on the VM.

B- The VM must be running an Ubuntu operating system.

C- A public IP address is necessary for Azure Machine Learning to communicate with the VM, enabling it to serve as a compute target.

D- The VM should possess a minimum hardware specification of 4 CPU cores.

E- The VM should conform to one of the general-purpose machine types recommended in Azure Machine Learning compute instance setup.

A- Removing nonessential code, like exploratory code and print statements, helps streamline the script for automation and reduces clutter.

B- Convert explanatory markdown cells from notebooks into code comments within the script.

C- Regularly test the script in an end-to-end automation pipeline to check its performance and output.

D- Refactoring code into smaller functions improves the script's structure, making it easier to read, debug, and maintain, which is crucial for automated production environments.

A- import mlflow; mlflow.log_all()

B- import mlflow; logger = mlflow(metrics="all", parameters="all", artifacts="all")

C- import mlflow; mlflow.autolog()

D- import mlflow; mlflow.log_metric("all=true")

A- Data wrangling is the process of transforming raw data into a structured and usable format that aligns with the requirements of a particular machine learning model or application. This process can include cleaning, structuring, and enriching raw data before it is used in analytic processes. It is more than just preprocessing; it often involves a more interactive and exploratory approach to understanding and transforming data.

B- Data wrangling involves dividing the dataset into training and testing subsets.

C- Data wrangling is synonymous with data preprocessing.

D- Data wrangling is the practice of managing, storing, and sharing data within a team of data scientists.

A- Split Data

B- Load Data

C- Join Data

D- Normalize Data

A- Recall

B- Precision

C- Confusion Matrix

D- Pearson Correlation value