Intro to Machine Learning Quiz

Which of the following best describes the process of fitting a model in the context of machine learning?

• Choosing a model that best fits the intuition of the analyst without the need for data.

• Dividing the dataset into two groups and then making predictions based on the historical average of each group.

• Using data to capture patterns and determine how to make predictions for new instances.

• Manually adjusting the parameters of a model to match predictions with real estate trends observed in the past.

Answer explanation

The process of fitting or training a model is described as capturing patterns from data to decide how to make predictions. This process involves using the training data to determine how to split the dataset and predict the new instances.

What advantage does a deeper decision tree have over a simpler one with fewer splits?

• It can make predictions without the need for any data.

• It simplifies the model by reducing the number of factors considered, making it easier to understand.

• It captures more factors affecting the prediction, potentially increasing accuracy.

• It requires less computational power to make predictions.

Answer explanation

While the simplest decision tree categorizes houses into two groups based on one factor, deeper trees with more splits can consider additional factors (like lot size, number of bathrooms, etc.), thus capturing more complexities in the data. This increased complexity can potentially make the model's predictions more accurate, as it reflects a wider range of factors affecting home prices.

When using Pandas to explore a dataset, what command is used to load data into a DataFrame from a CSV file?

• pd.load_csv('file_path')

• pd.read_csv('file_path')

• pd.import_csv('file_path')

• pd.open_csv('file_path')

Answer explanation

To load data from a CSV file into a Pandas DataFrame, the correct command is pd.read_csv('file_path'). This function is part of the Pandas library, which is typically imported and abbreviated as pd in Python code.

What does the "mean" value represent in the data description generated by Pandas?

• The most frequently occurring value in the dataset

• The middle value when the dataset is sorted in ascending order

• The average value of the dataset

• The dispersion of the values in the dataset

Answer explanation

In statistics, the mean represents the average value of a dataset. It is calculated by summing all the values in the dataset and then dividing by the number of observations. The mean is a measure of central tendency and provides insight into the typical value of the data.

What is the purpose of selecting a subset of features in a machine learning model?

• To increase the complexity of the model

• To decrease the accuracy of the model

• To reduce computational complexity and overfitting

• To introduce randomness in model training

Answer explanation

Selecting a subset of features helps in reducing the computational complexity and mitigating overfitting in machine learning models. Overfitting occurs when a model learns to capture noise in the training data rather than underlying patterns, which can lead to poor generalization on unseen data. By selecting only the most relevant features, the model becomes simpler and less prone to overfitting, resulting in better performance on new data.

Which of the following steps is NOT part of building and using a machine learning model?

• Defining the model type and parameters

• Fitting the model to the data

• Visualizing the data using matplotlib

• Evaluating the model's predictions

Answer explanation

While data visualization is an important aspect of exploratory data analysis and model interpretation, it is not a direct step in building and using a machine learning model. The key steps in building and using a machine learning model include defining the model type and parameters, fitting the model to the data, and evaluating the model's predictions. Data visualization helps in understanding the data distribution, relationships between variables, and identifying potential patterns, but it is not directly involved in the model building process.

What is the purpose of model validation in machine learning?

• To measure the complexity of the model

• To evaluate the model's predictive accuracy on new data

• To increase the number of features in the model

• To minimize the mean absolute error

Answer explanation

Model validation is essential in machine learning to assess how well a trained model generalizes to unseen data. The main goal of model validation is to evaluate the predictive accuracy of the model on new, unseen data points. This process helps in understanding how well the model performs in real-world scenarios and whether it can make accurate predictions beyond the training data.

What is the drawback of measuring predictive accuracy using "in-sample" scores?

• It requires additional computational resources

• It tends to underestimate the model's performance

• It leads to overfitting the model to the training data

• It may not reflect the model's performance on new data

Answer explanation

The drawback of measuring predictive accuracy using "in-sample" scores is that it may not reflect the model's performance on new data. Since the model has already seen the training data during the training phase, evaluating its performance on the same data can lead to overfitting and does not provide an accurate estimate of how well the model will generalize to unseen data.

What is the main drawback of a decision tree model with too few splits?

• It results in overfitting, where the model matches the training data perfectly but performs poorly on new data

• It results in underfitting, where the model fails to capture important patterns in the data and performs poorly on both training and validation data

• It increases the computational complexity of the model, making it slower to train

• It results in a more accurate model due to fewer splits and simpler structure

Answer explanation

A decision tree model with too few splits (i.e., shallow depth) tends to underfit the data. Underfitting occurs when the model is too simplistic to capture the underlying patterns in the data, leading to poor performance not only on the training data but also on new, unseen data (validation data). In underfitting, the model's predictions are generally inaccurate and fail to capture the complexity of the relationship between features and the target variable. Option A describes the issue of overfitting, where the model fits the training data too closely, but performs poorly on new data.

What is a key advantage of using a random forest model over a single decision tree?

• Random forests have a simpler structure, making them easier to interpret

• Random forests tend to overfit the data less than a single decision tree

• Random forests use fewer computational resources compared to a single decision tree

• Random forests generally have better predictive accuracy than a single decision tree

Answer explanation

The key advantage of using a random forest model over a single decision tree is that random forests generally have better predictive accuracy. Random forests combine multiple decision trees and make predictions by averaging the predictions of each component tree, resulting in improved performance. This ensemble approach helps reduce overfitting and captures more complex patterns in the data compared to a single decision tree.