Classical ML Time Series P2

Why is it important to split data into training and testing sets in machine learning?

• A) To reduce the overall dataset size and save memory

• B) To evaluate the model on data it hasn't seen before, helping to assess its ability to generalize

• C) To ensure the model has access to all data, maximizing its learning potential

• D) To make the training process faster by using only a subset of data

What is "overfitting" in the context of machine learning models?

• A) When a model performs well on the training data but poorly on new, unseen data

• B) When a model has too few features, making it underperform on all data

• C) When a model uses only a small portion of the data for training

• D) When a model is unable to capture any patterns in the data

Which of the following techniques can help mitigate overfitting in machine learning models?

• A) Reducing the size of the dataset

• B) Training the model for fewer epochs without any validation

• C) Using feature engineering, cross-validation, and hyperparameter tuning

• D) Increasing the model's complexity by adding more layers or features

What is cross-validation in the context of time series analysis?

• A) Randomly splitting the data into training and testing sets multiple times to evaluate model performance

• B) Shuffling the data and performing k-fold validation like in non-time series data

• C) Using only the most recent data points for testing and ignoring the earlier data points

• D) Sequentially splitting the data into training and testing sets over time to preserve temporal order

What is the main difference between a parameter and a hyperparameter in machine learning?

• A) Parameters are set before training, while hyperparameters are learned during training

• B) Hyperparameters are set before training, while parameters are learned during training

• C) Both parameters and hyperparameters are optimized automatically by the model

• D) Parameters and hyperparameters refer to the same values in a model

1. You have a dataset of customer transactions without any predefined categories. Which of the following approaches would be appropriate?

• Supervised Learning

• Unsupervised Learning

• Regression

• Classification

For predicting house prices, which of these statements are true?

• SML - Classification

• UML - Regression

• UML - Clustering

• SML - Regression

In spam email detection, which approaches could be valid?

• SML - Classification

• UML - Regression

• UML - Clustering

• SML - Regression