Passing exam with DP-100 latest training materials

DP-100 Exam Outline

The Microsoft DP-100 was recently renewed to meet the most current market needs and now it measures the following skills:

• Optimizing and Managing Models;
• Deploying and Consuming Models;
• Running Experiments and Training Models.
• Setting Up the Workspace for Azure Machine Learning;

The DP-100 exam domain of Setting Up the Workspace for Azure Machine Learning (ML) has three sections. The first touches on creating the workspace for ML. Here, you're to come across tasks like creating and configuring the workspace and managing it using Azure ML studio. The next part is concerning data object management within the workspace of Azure ML, where the focus goes to registering and maintaining datasets. The final aspect regards maintaining contexts for experiment compute. Under this, there will be creating instances for compute, determining the appropriate specs for compute targeting workload training, and developing targets for compute directed at experiments as well as training.

Regarding Optimizing and Managing Models, candidates will build their skills in five crucial areas. To begin is the area of creating optimal models using automated ML. This takes into account areas like Azure ML studio, Azure ML SDK, scaling options for pre-processing, algorithm determination, and getting data to be utilized in running the automated ML. The next thing goes into tuning hyperparameters using hyperdrive. Candidates need to note the sampling methods, search space, primary metric, termination options, and the right model. Another field concerns managing models where coverage includes model interpreters and feature importance data. Finally, students will learn how to manage models by exploring trained model registration, monitoring model usage, and monitoring data drift.

The Microsoft DP-100 exam also deals with the Deploying and Consuming Models. Of interest, there are four sections. It starts with the creation of targets for production compute involving security meant for deployed services & compute options targeting deployment. It's followed by the part of deploying a model as a service. This touches deployment settings, consuming deployed services, and troubleshooting issues for deployment containers. The next segment is creating a batch interference pipeline. Finally, students look at publishing a web service in the form of a designer pipeline. Issues also covered are compute resource, inference pipeline, and consumption of an already deployed endpoint.

The last DP-100 exam domain talks about Running Experiments and Training Models. The first way to achieve abilities in this area is by learning how to use Azure ML Designer to create models. This will be actualized by exploring creation of a training pipeline, ingestion of data within a designer pipeline, defining data flow for a pipeline using designer modules, and using modules for custom code. The second one regards running training scripts within the Azure ML workspace. Within this sphere, the students' focus will be how to use the Azure ML SDK in consuming data from a dataset in an experiment. The third thing in this topic has to do with using an experiment run to generate metrics. Here, learning includes log metrics, retrieving and viewing experiment outputs, and troubleshooting experiment errors using logs. The fourth and final area of concern is automating the process of model training. This includes developing a pipeline by utilizing the SDK, passing data, running a pipeline, and monitoring pipeline runs.

Reference: https://www.microsoft.com/en-us/learning/exam-dp-100.aspx

Microsoft DP-100 Exam Overview:

The Microsoft DP-100 test helps candidates check their machine learning knowledge and skills and prove themselves as qualified data scientists.

Microsoft DP-100 Exam Syllabus Topics: