Applying Machine Learning to your Data with GCP

Which of these is a TRUE statement?

* Machine Learning models learn the relationship between your dataset features and label on your training dataset to then predict on a dataset where the correct label is unknown. Learning models learn the relationship between your dataset labels and feature on your training dataset to then predict on a dataset where the correct instance is unknown. Machine Learning models learn the relationship between your dataset features and instances on your training dataset to then predict on a dataset where the correct instance is unknown.

What is an instance in a training dataset?

A single observation or record of data

What does a sentiment score of -1.0 at a magnitude of 1.0 represent for text that has been processed through the NLP API?

A strongly negative sentiment

Machine learning can be best described as: Select all 3 correct answers.

An iterative cycle of continual data research, model refinement, and insights through a learning feedback loop A potentially viable option if you need to perform predictive analysis like "What is my revenue going to be next quarter?" A method of replacing heuristic rules and hardcoded logic trees with learning models

If you find out that your code cell isn't executing, what is a good step to check?

Be sure the cell you're in is a code cell and not a markdown cell Ensure you have the correct permissions to the underlying operation (e.g. BigQuery dataset, GCS bucket) You may have lost connection to your datalab instance, consider trying datalab connect [name] inside of Cloud Shell * All of the above are good troubleshooting options

Which of these are best practices when working with Cloud Datalab notebooks? Select all 2 correct answers.

Begin your notebooks with markdown cells explaining the purpose and goal of the notebook You can query your datasets directly from within Cloud Datalab to BigQuery to show steps in your data analysis

What is the tool within Google Cloud Platform that launches your Cloud Datalab notebook instance on Google Compute Engine?

Cloud Shell

True or False: Because of the learning nature of ML Models, you will spend very little time cleaning and processing your datasets

False

True or False: You need to always alias your BigQuery library with "bq" or else it won't be recognised in your code cells

False

When creating a dataset for use in ML training, which of the below are the best practices to keep in mind?

For faster iteration, consider training on just a sample of your dataset

What is one of the core benefits of having SQL code inside of a Cloud Datalab notebook?

It will always execute faster than running the same query through the BigQuery Web UI You have access to the same BigQuery commands and the query validator icon just like the Web UI * You can breakup a SQL experimental notebook with markdown and only run portions of your code at a time All of the above

Which critical step is missing from the below model phases? Ingest data, ? Build a Training Dataset, Create a Model, Evaluate your Model, Make Predictions

Preprocess Data

What is a key input to a pre-trained ML model like the Translation API microservice? Select all 2 correct answers.

You need to provide the model with expected parameters like which languages you are translating to and from You will need to provide the model with a data input -- for translation this is normally a block of text