Intro to Machine Learning Concepts (Day 1-Test 1)

What are the key steps in data collection and preparation?

• Create visualizations, conduct interviews, publish papers

• Define objectives, identify data sources, collect data, clean data, transform data, validate data.

• Develop software, implement algorithms, deploy systems

• Analyze results, report findings, archive data

How does supervised learning differ from unsupervised learning?

• Supervised learning requires no data, while unsupervised learning requires a lot of data.

• Supervised learning is used for clustering, while unsupervised learning is used for classification.

• Supervised learning uses labeled data, while unsupervised learning uses unlabeled data.

• Supervised learning is faster than unsupervised learning in all cases.

What is the purpose of model training in machine learning?

• To create a static model that does not change over time.

• The purpose of model training is to enable the model to learn from data and improve its predictive accuracy.

• To eliminate the need for data preprocessing.

• To reduce the amount of data used in predictions.

Explain the significance of a confusion matrix in evaluating model performance.

• The confusion matrix is significant as it provides a comprehensive view of model performance, enabling the calculation of key metrics and insights into prediction errors.

• It is used to visualize the training data.

• It only shows the accuracy of the model.

• It provides a single metric for model performance.

What are common performance measures used to assess a classification model?

• R-squared

• Log Loss

• Accuracy, Precision, Recall, F1 Score, AUC-ROC

• Mean Squared Error

Describe the process of model deployment in a machine learning project.

• Data collection and preprocessing only

• Model training without evaluation

• Directly deploying without testing

• The process of model deployment in a machine learning project includes training, evaluation, serialization, environment setup, API development, deployment, and monitoring.

What is the role of maintenance in the machine learning lifecycle?

• Maintenance is focused solely on data collection and not on model performance.

• Maintenance ensures the longevity and effectiveness of machine learning models by adapting to changes and improving performance.

• Maintenance involves creating new models from scratch every time.

• Maintenance is only necessary during the initial model training phase.

What techniques are used for data preprocessing before model training?

• Data cleaning, transformation, feature selection, and encoding categorical variables.

• Hyperparameter tuning methods

• Model evaluation metrics

• Data visualization techniques

What are the main types of unsupervised learning algorithms?

• Supervised Learning and Regression

• Clustering and Dimensionality Reduction

• Neural Networks and Decision Trees

• Classification and Reinforcement Learning

What is the importance of feature engineering in machine learning?

• Feature engineering is the same as data collection.

• Feature engineering only applies to supervised learning.

• Feature engineering helps improve model performance by creating new input features from existing data.

• Feature engineering is unnecessary if you have a large dataset.