Analytics that answers **"Why did it happen?"** by looking for reasons and relationships in past data, often using comparisons and correlations.

Analytics that answers **"What is likely to happen?"** by using past data (often with ML) to estimate future outcomes.

Using data, statistical methods, and sometimes ML to understand what is happening and support decisions in organizations.

What is Data science?

An interdisciplinary field that combines statistics, computing, and domain knowledge to collect, clean, analyze, model, and communicate data‑driven insights.

What is Generative AI?

A type of AI system that can create new content (such as text, images, or code) by learning patterns from large datasets.

What is Supervised learning?

Machine learning that uses labeled examples (inputs with known correct outputs) to learn a mapping from inputs to outputs.

Analytics vs Machine Learning vs AI: Cutting Through the Buzzwords — Data Science Foundations: From Raw Data to Insight

Q: A bank uses past customer data (age, income, past payments, etc.) and whether each person actually repaid a loan to train a model that scores new loan applicants as low, medium, or high risk. What best describes this activity?

Predictive analytics with supervised learning. The bank is using labeled past data (repaid or not) to train a model that predicts risk for new applicants. That is **supervised learning**, and the goal is to estimate a **future outcome**, which is **predictive analytics**.

Q: Descriptive analytics

Analytics that answers **"What happened?"** by summarizing past data using counts, averages, percentages, and charts.

Q: Prescriptive analytics

Analytics that answers **"What should we do?"** by recommending actions or decisions based on predictions, goals, and constraints.

Q: Machine learning (ML)

A set of methods that **learn patterns from data** to make predictions or decisions, instead of relying only on hand‑written rules.

Big Picture: Analytics, ML, and AI

Three Buzzwords

You will untangle three common buzzwords: analytics, machine learning (ML), and artificial intelligence (AI), and connect them to descriptive statistics and probability.

Simple Definitions

Analytics: Using data to understand what is happening and what to do.
ML: Algorithms that learn patterns from data to make predictions or decisions.
AI: Systems that do tasks linked to human-like intelligence.

How They Relate

Think of AI as the big umbrella, machine learning as one major part under it, and analytics as a broad practice that overlaps with both but also includes classic, non-ML methods.

Why It Feels Confusing

Since the early 2010s and especially by 2026, many companies label almost everything as "AI". In reality, much of it is basic analytics or standard ML, not sci‑fi robots.

What You Will Learn

You will break analytics into four levels, understand supervised vs unsupervised learning at a high level, and see how ML fits inside data science and AI with everyday examples.

The Four Types of Analytics

Four Types of Analytics

Analytics is often split into four types: descriptive, diagnostic, predictive, prescriptive. They build on each other and connect to descriptive statistics and probability.

Descriptive Analytics

Descriptive: What happened? It summarizes past data with counts, averages, percentages, and charts, like a dashboard showing hours watched last week.

Diagnostic Analytics

Diagnostic: Why did it happen? It looks for reasons and relationships, using comparisons and correlations, like finding that watch time dropped because a show ended.

Predictive Analytics

Predictive: What is likely to happen? It uses past data and probability (often with ML) to estimate future outcomes, like predicting how many users will cancel next month.

Prescriptive Analytics

Prescriptive: What should we do? It suggests actions based on predictions and goals, using rules or optimization, like deciding which users to send discount offers to.

The Analytics Ladder

Think of a ladder: descriptive (what), diagnostic (why), predictive (what next), prescriptive (what to do). ML mainly powers predictive and prescriptive, but all four matter in practice.

Classify the Analytics Type

Try this quick thought exercise. For each scenario, decide if it is descriptive, diagnostic, predictive, or prescriptive analytics.

A fitness app shows you a bar chart of how many steps you took each day last week.
A university compares grades before and after switching to online exams to see if scores changed.
An online store estimates how many units of a product it will sell next month.
A ride‑sharing app suggests higher prices in certain areas at certain times to balance supply and demand.

Your turn:

Write down your answers in the format: `1: type, 2: type, 3: type, 4: type`.

Then check yourself:

1: Descriptive (just summarizing what happened)
2: Diagnostic (trying to understand why scores changed)
3: Predictive (estimating future sales)
4: Prescriptive (recommending actions: change prices)

If any felt tricky, reread the four definitions and think: Am I looking at past facts, reasons, future outcomes, or recommended actions?

What Is Machine Learning, Really?

Plain Definition of ML

Machine learning is about learning patterns from data so a system can make predictions or decisions without humans writing all the detailed rules.

Classic vs ML Approach

Classic programming uses hand‑written rules like "if age < 25 and income < 20k then X". ML instead learns patterns from many examples of inputs and outcomes.

ML and Probability

ML is usually probabilistic, giving predictions with uncertainty, not perfect answers. This links directly to the probability and uncertainty ideas you learned earlier.

ML Inside Data Science

ML is one part of data science, which also covers collecting data, cleaning it, visualizing it, and explaining results to people who make decisions.

Everyday ML Examples

Examples include spam filters, streaming recommendations, phone face unlock, and fraud alerts. These are predictive analytics tasks powered by ML models.

Supervised vs Unsupervised Learning (No Math)

Two Big ML Families

Most ML methods are either supervised or unsupervised. You can understand the difference without equations by focusing on whether you have labels.

Supervised Learning

In supervised learning, each example has an input and a correct answer (label). The model learns to map inputs to outputs, like predicting house prices or spam vs not‑spam.

Studying With an Answer Key

Supervised learning is like studying with an answer key: you see many question–answer pairs and learn patterns, then answer new questions on your own.

Unsupervised Learning

In unsupervised learning, you only have inputs, no labels. The model looks for patterns or groups, like clustering customers into segments.

Exploring Without a Guide

Unsupervised learning is like exploring a new city alone and noticing natural clusters (shopping area, quiet streets) without anyone giving official labels.

Link to Analytics

Supervised learning mostly powers predictive analytics, while unsupervised learning helps in descriptive/diagnostic work by revealing hidden structure.

Mini Case: Streaming Service Decisions

Streaming Service Scenario

Imagine a streaming service trying to reduce user churn (cancellations). This scenario shows how analytics, ML, and AI work together in a realistic setting.

Descriptive Step

First, descriptive analytics: dashboards show total users, cancellations, and average watch time by month. They notice cancellations spike in March.

Diagnostic Step

Next, diagnostic analytics: they compare March to earlier months and see fewer new episodes, more app bugs, and lower watch time among users who canceled.

Predictive Step with ML

Then, predictive analytics using supervised ML: a model learns from past users to predict who is likely to cancel next month based on behavior and history.

Prescriptive Step

Finally, prescriptive analytics: they use predictions to decide actions like sending offers to high‑risk users or optimizing which users get which discount.

Where AI Appears

Advanced recommendation engines or chatbots that answer questions in natural language are marketed as "AI", but they are powered by ML models that support analytics.

Quick Check: Analytics Types and ML

Answer this question to check your understanding of analytics types and ML.

A bank uses past customer data (age, income, past payments, etc.) and whether each person actually repaid a loan to train a model that scores new loan applicants as low, medium, or high risk. What best describes this activity?

Descriptive analytics with unsupervised learning
Predictive analytics with supervised learning
Prescriptive analytics without machine learning
Diagnostic analytics with supervised learning

Show Answer

Answer: B) Predictive analytics with supervised learning

The bank is using labeled past data (repaid or not) to train a model that predicts risk for new applicants. That is **supervised learning**, and the goal is to estimate a **future outcome**, which is **predictive analytics**.

Where Does AI Fit In?

Modern View of AI

AI is about building systems that do tasks linked to human intelligence, like understanding language or recognizing images. Today, most practical AI uses machine learning.

Examples of AI Systems

Examples include generative AI that writes text, image recognition that finds objects, speech‑to‑text systems, and game‑playing agents that learn complex strategies.

AI vs ML vs Analytics

Many AI applications are built with ML models. Many "AI" features in business are just predictive or prescriptive analytics powered by ML.

Analytics Without AI

Descriptive and diagnostic analytics often use only statistics and visualization, with no ML or AI. Dashboards and summary reports are classic examples.

Ask Better Questions

When you hear "we use AI", ask: which analytics type is this, are they using ML, and is it supervised or unsupervised? Or is "AI" just a marketing label?

Spot the Buzzword vs Reality

Practice cutting through hype. For each statement, decide what is actually going on.

"Our AI dashboard shows real‑time sales and compares them to last week."

Likely reality: Descriptive analytics (maybe some diagnostic) with no real AI.

"We use AI to predict which students are at risk of failing and alert advisors."

Likely reality: Predictive analytics using supervised ML.

"Our AI segments customers into groups based on behavior."

Likely reality: Diagnostic/descriptive analytics using unsupervised ML (clustering).

"We have an AI assistant that answers customer questions in chat."

Likely reality: AI application powered by ML models (often large language models), plus analytics in the background.

Your task:

Pick a product or service you use (social media app, shopping site, banking app, etc.).
Write down one feature they might call "AI".
Decide:

Which analytics type is it mainly (descriptive, diagnostic, predictive, prescriptive)?
Is it likely using ML? If yes, supervised or unsupervised?

This short reflection will help you apply these ideas whenever you see "AI" in the news or in marketing.

Key Term Review

Use these flashcards to review the core terms from this module.

Descriptive analytics: Analytics that answers **"What happened?"** by summarizing past data using counts, averages, percentages, and charts.
Diagnostic analytics: Analytics that answers **"Why did it happen?"** by looking for reasons and relationships in past data, often using comparisons and correlations.
Predictive analytics: Analytics that answers **"What is likely to happen?"** by using past data (often with ML) to estimate future outcomes.
Prescriptive analytics: Analytics that answers **"What should we do?"** by recommending actions or decisions based on predictions, goals, and constraints.
Machine learning (ML): A set of methods that **learn patterns from data** to make predictions or decisions, instead of relying only on hand‑written rules.
Supervised learning: A type of ML that learns from **labeled examples**, where each input has a known correct answer (label), to predict labels for new inputs.
Unsupervised learning: A type of ML that works with **unlabeled data**, finding patterns, groups, or structure without given correct answers.
Artificial intelligence (AI): The broader field of building systems that perform tasks linked to human intelligence, such as understanding language, recognizing images, or planning.
Data science: A practice that combines data collection, cleaning, analysis, visualization, and modeling (including ML) to extract insights and support decisions.
Generative AI: AI systems (often large ML models) that can **generate** new content such as text, images, code, or audio, based on patterns learned from data.

Key Terms

Analytics: Using data, statistical methods, and sometimes ML to understand what is happening and support decisions in organizations.
Data science: An interdisciplinary field that combines statistics, computing, and domain knowledge to collect, clean, analyze, model, and communicate data‑driven insights.
Generative AI: A type of AI system that can create new content (such as text, images, or code) by learning patterns from large datasets.
Supervised learning: Machine learning that uses labeled examples (inputs with known correct outputs) to learn a mapping from inputs to outputs.
Diagnostic analytics: Analytics focused on understanding causes and relationships in past data to answer "Why did it happen?".
Predictive analytics: Analytics that uses past data to estimate future outcomes and answer "What is likely to happen?".
Descriptive analytics: Analytics focused on summarizing and describing past data to answer "What happened?".
Machine learning (ML): A collection of algorithms that learn patterns from data to make predictions or decisions without being explicitly programmed with all the rules.
Unsupervised learning: Machine learning that uses unlabeled data to find patterns, groups, or structure without known correct outputs.
Prescriptive analytics: Analytics that recommends actions or decisions based on predictions, goals, and constraints, answering "What should we do?".
Artificial intelligence (AI): The broader field of creating systems that can perform tasks that usually require human intelligence, such as perception, language understanding, and problem‑solving.