Vocabulary Flashcards – Sentiment Analysis of U.S. Import Tariffs Thesis

A set of vocabulary flashcards covering essential terms and techniques mentioned in the lecture notes on comparative sentiment-analysis research using Naïve Bayes and Logistic Regression.

Sentiment Analysis

The process of automatically identifying and categorising opinions in text as positive, negative or neutral.

Logistic Regression

A supervised learning algorithm that models the probability of a categorical outcome using the logistic (sigmoid) function.

Naïve Bayes

A probabilistic classifier that assumes feature independence and applies Bayes’ theorem to predict class membership.

VADER (Valence Aware Dictionary and sEntiment Reasoner)

A lexicon- and rule-based tool tailored for social-media text that assigns polarity scores and compound sentiment values.

SMOTE (Synthetic Minority Over-sampling Technique)

A resampling method that creates synthetic examples for minority classes to balance imbalanced datasets.

TF-IDF (Term Frequency–Inverse Document Frequency)

A weighting scheme that reflects how important a word is to a document relative to a corpus.

Tokenization

The preprocessing step that splits raw text into smaller units such as words or sub-words called tokens.

Stemming

Reducing words to their root or base form to unify word variants (e.g., ‘running’ → ‘run’).

Stopword Removal

Eliminating very common words (e.g., ‘and’, ‘the’) that carry little semantic value in text analysis.

CRISP-DM

A six-phase, industry-standard methodology for data-mining projects: Business Understanding, Data Understanding, Data Preparation, Modelling, Evaluation and Deployment.

Machine Learning

A branch of artificial intelligence that enables systems to learn patterns from data and improve performance over time.

Supervised Learning

A machine-learning paradigm where models are trained using labelled input–output pairs.

GridSearchCV

An exhaustive search technique in scikit-learn that tests multiple hyperparameter combinations using cross-validation.

Precision

The proportion of true positive predictions among all positive predictions made by a model.

Recall

The proportion of true positive predictions captured out of all actual positive instances.

F1-Score

The harmonic mean of precision and recall; balances both metrics into a single measure.

Confusion Matrix

A table showing correct and incorrect predictions broken down by each class, used to evaluate classification models.

WordCloud

A visual representation where word size indicates frequency, revealing prominent terms in text data.

Decision Boundary

The surface that separates different class regions in the feature space according to a classifier.

Uji McNemar (McNemar Test)

A non-parametric statistical test for paired nominal data used to compare two classifiers on the same samples.

API X (formerly Twitter API)

An interface that allows programmatic access to X/Twitter data for retrieving, posting or analysing tweets.

Google Colab

A cloud-based Jupyter Notebook environment providing free CPU, GPU and TPU resources for Python code execution.

TruncatedSVD

A dimensionality-reduction technique that projects high-dimensional sparse data (e.g., TF-IDF) into lower dimensions.

One-vs-Rest Strategy

A multi-class classification approach that trains one binary classifier per class against all other classes.

Vectorizer

A tool (e.g., CountVectorizer, TfidfVectorizer) that converts text into numerical feature vectors for machine-learning models.

Lexicon-Based Analysis

Sentiment detection that relies on predefined dictionaries of words annotated with polarity scores.

Class Imbalance

A situation where some classes have far fewer samples than others, often degrading model performance.

Hyperparameter Tuning

The process of optimising external configuration settings (e.g., C value in Logistic Regression) to improve model performance.