Ultimate Engineering Study Guide - Questions & Answers

Transcribed image text: When is a task considered to be "unsupervised"? O A task is unsupervised when you are using labeled data. O A task is unsupervised when you are using unlabeled data. A task is unsupervised when you define a reward function. O All of the above. An application that uses data about homes and corresponding labels to predict home sale prices uses what kind of machine learning? O Supervised Unsupervised Reinforcement learning O All of the above An application that uses data about homes and corresponding labels to predict home sale prices uses what kind of machine learning? O Supervised Unsupervised Reinforcement learning All of the above Which of the following is not a reason why it is important to inspect your dataset before training a model? Data needs to be transformed or preprocessed so it's in the correct format to be used by your model Machine learning handles all of the reasoning about data for you. Understanding the shape and structure of your data can help you make more informed decisions on selecting a model. You can find missing or incomplete values. When checking the quality of your data, what should you look out for? Outliers Categorical labels O Training algorithms O All of the above What is the definition of model accuracy? O How often your model makes a correct prediction. How often your model makes similar predictions. How well the results mimic a specific shape of an algorithm. Does the prediction reflect reality. Which of the following is not a model evaluation metric? O Root Mean Square (RMS) Model Inference Algorithm Silhouette Coefficient O Accuracy Which of the following is only a characteristic of reinforcement learning? O Uses labels for training data. Does not use labels for training data. Uses a reward function. O All of the above. You are creating a program to identify dogs using supervised learning. What is not an example of a categorical label? Is a dog. O is not a dog. O May be a wolf. All of the above. In reinforcement learning, the agent: Receives reward signals from the environment for its actions. Is a piece of software you train to learn by interacting with an environment. Has a goal of maximizing its total reward over time. O All of the above. What are hyperparameters? Model parameters that change faster than most other model parameters during model training. Model parameters that have more of an impact on the final result than most other model parameters. Parameters within a model inference algorithm. O Parameters which affect model training but typically cannot be incrementally adjusted during training like other parameters. True or False: As part of building a good dataset you should use data visualizations to check for outliers and trends in your data. True False