D491 Introduction to Analytics - Set 4 - Part 2

Correct Answer: The error due to simplifying assumptions in the model

Bias refers to the error introduced by approximating complex real-world problems with simplified models, often leading to underfitting.

Correct Answer: One-hot encoding

One-hot encoding transforms categorical variables into binary vectors, where each category is represented as a unique combination of 0s and 1s.

Correct Answer: ARIMA

ARIMA (AutoRegressive Integrated Moving Average) is a popular model used for time series forecasting by capturing trends, seasonality, and noise.

Correct Answer: Louvain method

The Louvain method is a popular algorithm for detecting communities in large networks by optimizing modularity at multiple levels.

Correct Answer: It includes regularization to prevent overfitting

XGBoost includes L1 and L2 regularization to prevent overfitting, making it more robust than traditional gradient boosting methods.

Correct Answer: To prevent overfitting by randomly dropping units during training

Dropout randomly "drops" units (nodes) from the network during training to prevent the model from becoming too reliant on specific pathways.

Correct Answer: Down-sampling the majority class

Down-sampling reduces the number of instances in the majority class to balance the dataset, making it more suitable for classification tasks.

Correct Answer: Dropout

Dropout is a technique that randomly drops units during training to prevent overfitting and improve the generalization ability of the model.

Correct Answer: The problem where increasing data dimensions makes clustering harder

The curse of dimensionality refers to the phenomenon where the performance of machine learning algorithms decreases as the number of input dimensions increases.

Correct Answer: To identify unusual patterns in the data

Anomaly detection algorithms are designed to identify data points that deviate significantly from the rest of the data, indicating possible errors or fraud.

Correct Answer: Recurrent Neural Networks (RNNs)

RNNs are suitable for NLP tasks because they are designed to handle sequential data, such as text, by using previous inputs to inform current predictions.

Correct Answer: They are prone to overfitting

Decision trees are prone to overfitting, especially when the tree becomes too complex by learning details specific to the training data.

Correct Answer: Continuous

A box plot visualizes the distribution of continuous data by displaying the median, quartiles, and possible outliers.

Correct Answer: Gradient descent

Gradient descent is an optimization algorithm used to minimize the loss function by iteratively adjusting model parameters in the direction of the steepest descent.

Correct Answer: Naive Bayes

Naive Bayes is commonly used for spam detection because of its ability to handle large-scale categorical data and probabilistic classification.

Correct Answer: A model that performs well on the training data but poorly on test data

Overfitting occurs when a model becomes too complex, capturing noise and specific patterns in the training data that do not generalize to unseen data.

Correct Answer: Imputation

Imputation is the process of filling in missing data points with estimated values, such as the mean, median, or mode of the dataset.

Correct Answer: Decision tree

Decision trees use a series of "if-then" rules to classify data points by splitting features into different categories at each node.

Correct Answer: Supervised learning uses labeled data, while unsupervised learning uses unlabeled data

Supervised learning algorithms use labeled data to make predictions, while unsupervised learning algorithms identify patterns in data without labeled outcomes.

Correct Answer: Diagnostic analytics

Diagnostic analytics focuses on identifying the root cause of patterns or issues by analyzing historical data and uncovering trends.