FBLA - Data Science & AI

Data Science

The field that uses data, stats, and computation to extract insights and support decisions.

Data Conversion

Changing data from one numerical system or format to another (binary ↔ decimal ↔ hex)

Dependence Methods

Statistical techniques showing how variables rely/depend

on each other.

Interdependence Methods

Techniques examining relationships without a dependent variable. (Inter = Without)

Machine Learning Algorithm

A procedure computers use to learn patterns from data.

Learning Function

The rule a machine learning model uses to update itself based on errors.

Training Dataset

Data used to teach a model

Validation Dataset

Data used to fine-tune a model during training.

Test Dataset

Data used to measure final model performance.

Deep Learning

Machine learning using layered neural networks to learn complex patterns.

Structured Data

Data organized in rows and columns (like spreadsheets and SQL tables).

Unstructured Data

Data without a fixed format, such as text, images, audio, or video.

Numeric Data

Data measured or counted using numbers.

Categorical Data

Data grouped by labels or categories.

Binary System

Base-2 number system using 0 and 1.

Decimal System

Base-10 number system using digits 0-9.

Hexadecimal System

Base-16 system using 0-9 and A-F, often used in computing.

Data Sources

Places where data originates/comes from (the source)

(ex. sensors, surveys, transactions, social media).

Data Wrangling

Cleaning and organizing raw data so it can be analyzed. (Wrangling = cleaning/organizing)

Data Transformation

Converting/transforming data into a more useful or usable structure.

Data Science Process

Ask questions → collect → clean → analyze → model → interpret → communicate.

defining the problem, collecting and cleaning data, exploring and analyzing it to find patterns, modeling and evaluating solutions, and finally communicating the results

Probability

The chance an event happens; found by favorable outcomes ÷ total outcomes.

Sample Space

All possible outcomes in a scenario.

Random Variable

A variable whose value depends on random events.

Mean

Average of a dataset; add values and divide by number of values.

Median

Middle value when data is sorted; if even count, average the two middle numbers.

Mode

Value that appears most frequently.

Range

Largest value minus smallest value.

Variance

Measures spread of data from the mean; standard deviation squared.

Standard Deviation

Shows how spread out data is from the mean; square root of variance.

Expected Value

Long-term average outcome of a random variable.

Correlation

How strongly two variables move together (positive, negative, or none).

Regression

Predicting outcomes using relationships in data (line of best fit).

Outlier

Data point far from others that can distort results.

Gaussian (Normal) Distribution

Bell-shaped curve where values cluster around the mean.

Covariance

Shows whether two variables increase or decrease together.

Discrete Variable

A variable with countable values (like # of students).

Continuous Variable

A variable with infinite possible values within a range (height, time).

Data Visualization

Displaying data visually to reveal trends or patterns.

Bar Graph

Used to compare categories.

Histogram

Shows distribution of numerical data.

Line Graph

Tracks trends or changes over time.

Scatter Plot

Shows relationships or correlations between two numeric variables.

Box Plot

Shows spread, quartiles, and outliers in large datasets.

Pie Chart

Shows parts of a whole as percentages.

Heatmap

Shows intensity or correlations using color patterns.

Multivariate Data

Data involving more than two variables.

Linear Regression

Predicts a numeric value using a best-fit line.

Multiple Linear Regression

Predicts outcomes using several variables.

Logistic Regression

Predicts categories (yes/no, pass/fail).

(logical statement, like true/false)

Data Cleaning

Fixing errors, removing duplicates, and handling missing values.

Data Quality Issues

Problems like incomplete data, errors, duplicates, or noise.

K-Means

Clustering algorithm that groups data by similarity.

Decision Tree

Model that predicts outcomes by splitting data into branches.

Python

Main programming language for AI and data science due to simplicity and powerful libraries.

NumPy

Library for fast math and array operations.

Pandas

Library for cleaning, analyzing, and organizing data in dataframes.

Matplotlib

Library for visualizing data with graphs.

(mat = math = data) (plot = graph)

Seaborn

Python visualization library built on top of Matplotlib for cleaner graphs.

TensorFlow

Framework for building and training neural networks.

PyTorch

Deep learning framework popular for research and model development.

SQL

Language for storing, querying, and managing data in databases.

Jupyter Notebook

Environment to write code, visualize results, and explain analysis.

Relational Database

Database using tables linked by relationships (keys).

Google Cloud

Platform for running AI models, storing data, and deploying applications.

IBM Cloud

Cloud platform offering AI tools, storage, and data processing services.

R

Programming language used for statistics and data visualization.

Artificial Intelligence

Systems that mimic human intelligence to perform tasks like decision-making and pattern recognition.

Generative AI

AI that creates content such as text, images, and code.

Limitations of Generative AI

Can hallucinate, make false claims, and repeat bias.

Uses of Generative AI

Chatbots, creative tools, support in healthcare, summarization, coding help.

Computer Vision

AI that interprets visual data from images or videos.

Natural Language Processing (NLP)

AI that understands and generates human language.

Speech Recognition

Converts spoken words into text.

Robotics

AI used in machines that interact with the physical world.

Large Language Models (LLMs)

Neural networks trained on massive text datasets for reasoning and generating responses.

(Ex. Chatgpt)

LLM Capabilities

Summarization, question answering, reasoning, code generation.

Perception

AI converting raw data (images, audio, text) into useful information.

Knowledge Representation

How AI stores facts and relationships internally.

Reasoning

AI using stored knowledge to make decisions and predictions.

Inference

AI reaching conclusions without being explicitly told.

Knowledge Graph

Web of connected facts showing relationships between things.

Symbolic Reasoning

Solving problems using rules and logic.

Predicate Logic

Logical statements used to express facts and relationships.

Bayesian Network

Graph showing probabilistic relationships between variables.

Directed Acyclic Graph (DAG)

Graph used in Bayesian networks that has no loops.

Data Privacy

Protecting personal data from misuse.

Data Security

Safeguarding data from unauthorized access or attacks.

Ethical AI

Designing AI that is fair, transparent, and does not harm people.

Algorithmic Bias

When AI produces unfair outcomes due to biased data.

Transparency

Explaining how an AI system makes decisions.

Accountability

Identifying who is responsible for AI outcomes.

Consent

Users must agree before their data is used.

AI Surveillance

AI systems that monitor people, raising privacy concerns.

Security Risks of LLMs

Data leaks, unauthorized training, privacy issues.

LLM Credibility Issues

Hallucinations, misinformation, false confidence.

AI Dilemmas

Ethical problems caused by AI decisions (e.g., self-driving choices).

Logic-Based Reasoning

Uses defined rules to reach conclusions.

Probability-Based Reasoning

Uses likelihood and uncertainty instead of strict rules.