Certification 70-774

Perform Cloud Data Science with Azure Machine Learning

Perform Cloud Data Science with Azure Machine Learning

What is the official description for this exam?

Audience Profile:
Candidates for this exam are data scientists or analysts who leverage azure cloud services to build and deploy intelligent solutions. Candidates for this exam will have a good understanding of Azure data services and machine learning, familiar with common data science processes such as filtering and transforming data sets, model estimation, and model evaluation.
Candidates for this exam will have experience in publishing an effective API for knowledge intelligence.

Preparation Materials:
Course Sequence to cover most material on the exam:

Please note that the questions may test on, but will not be limited to, the topics described in the bulleted text. Also note that this guide is not an official curriculum but rather should be viewed as a community resource.

Do you have feedback about the relevance of the skills measured on this exam? Please send Microsoft your comments . All feedback will be reviewed and incorporated as appropriate while still maintaining the validity and reliability of the certification process. Note that Microsoft will not respond directly to your feedback. We appreciate your input in ensuring the quality of the Microsoft Certification program.

If you have concerns about specific questions on this exam, please submit an exam challenge.

If you have other questions or feedback about Microsoft Certification exams or about the certification program, registration, or promotions, please contact your Regional Service Center.

Skills Measured

The following list contains the exam objectives and resources to prepare.

Documentation and Learning Paths:
Prepare data to be analyzed in Azure Machine Learning and export from AML
Import and Export data to Azure Machine Learning [1]
  1. Import and export data from Azure Blob storage [1] [2]
  2. Import and export data from Azure SQL [1] [2]
  3. Import and export data via Hive Queries [1]
  4. Import data from a website [1] [2] [3]
  5. Import from on-premises SQL [1]
Explore and summarize data
  1. Create univariate summaries [1] https://msdn.microsoft.com/en-us/library/azure/dn905936.aspx
  2. Create multivariate summaries, NOTE: cover group by statements [1] [2]
  3. Visualize univariate distributions [1]
  4. Leverage existing R or Python notebooks for custom summaries
  5. Leverage existing R or Python notebooks for custom visualizations
  6. Leverage zip archives to import external packages for R or Python [Python] [R]
Cleanse data for Azure Machine Learning
  1. Apply filters to limit a dataset to the desired rows [1] [2] [3]
  2. Identify and address missing data [1] [2]
  3. Identify and address outliers [1]
  4. Remove columns and rows of datasets [1] [2] [3] [4]
Transform and perform feature engineering
  1. Merge multiple datasets (by rows or columns) into a single dataset by columns [1]
  2. Merge multiple datasets (by rows or columns) into a single dataset by rows [1 (a join on a column shown in this video w/ two datasets)]
  3. Add columns that are combinations other columns [1 (w/ Apply Math Module – performed calc and appended col, but could also replace inline)]
  4. Manually select and construct features for model estimation [1]
  5. Automatically select and construct features for model estimation [1]
  6. Reduce dimensions of data through Perform Principal Component Analysis [1] [2]
  7. Manage variable metadata [1]
  8. Select standardized variable based on planned analysis
Develop machine learning models / Select appropriate algorithm or method [1] [2] [3]
  1. Select appropriate algorithm for predicting continuous label data [1]
  2. Select appropriate algorithm for supervised vs unsupervised scenarios [1]
  3. Identify when to select R or Python notebooks are appropriate
  4. Identify algorithm for grouping unlabeled data [1]
  5. Identify algorithm for classifying label data [1]
  6. Select appropriate ensemble
Initialize and Train Appropriate Models
  1. Tune hyperparameters manually [1]
  2. Tune hyperparameters automatically [1] [2]
  3. Split Data into training and testing datasets, including leveraging routines for cross-validation [1] [2]
  4. Build ensemble using stacking method
Validate models
Operationalize and manage Azure Machine Learning services
Deploy models using Azure ML
  1. Publish a model developed inside AML [1] [2]
  2. Publish an externally developed scoring function using an AML package [1]
  3. Web service parameters [1]
  4. Create and publish a recommendation model [1] [2]
  5. Create and publish a language understanding model
Manage Azure Machine Learning Projects and workspaces
  1. Create projects and experiments (no links for projects) [1]
  2. Add assets to a project
  3. Create new workspaces [1]
  4. Invite users to a workspace [1]
  5. Switch between different workspaces [1]
  6. Create a Jupiter notebook that references an intermediate dataset [1] [2]
Consume Azure ML models
  1. Connect to a published AML web service [1]
  2. Consume a published ML model programmatically using a batch execution service [1] [2]
  3. Consume a published ML model programmatically using a request response service [1]
  4. Interact with a published ML model using Excel [1]
  5. Publish models to the marketplace ??? [1]
Consume exemplar cognitive services APIs
  1. Consume Vision APIs to process images [1]
  2. Consume Language APIs to process text [1] [2]
  3. Consume Knowledge APIs to create recommendations [1]
Leverage other services for Machine Learning
Build and use neural networks with CNTK [1]
  1. Leverage N-series VMs for GPU acceleration [1]
  2. Build and train a 3-layer feed forward neural network [1]
  3. Determine when to implement a neural network
Leverage existing resources to streamline development
  1. Clone template experiments from Cortana Intelligence Gallery [1]
  2. Leverage Cortana Intelligence Quickstart to deploy resources [1]
  3. Leverage data science VM for streamlined development [1] [2]
Leverage HDInsights for data sciences at scale
  1. Appropriate type of HDI cluster [1]
  2. Perform Exploratory Data Analysis by using Spark SQL [1]
  3. Build and use Machine Learning models with Spark on HDI [1] [2] [3]
  4. Build and use Machine Learning using Map-reduce [1]
  5. Build and use Machine Learning using Microsoft R Server [1]
Leverage SQL Server R Services on Azure for database analytics [1]
  1. Deploy a SQL Server 2016 Azure VM [1]
  2. Configure SQL Server to allow execution of R scripts [1] [2]
  3. Execute R scripts inside T-SQL statements [1]