← Data Analytics from Scratch: SQL, Spreadsheets and Metrics
Lesson
Why analytics, and the four kinds of analysis
Learner can explain what a data analyst actually does and tell apart descriptive, diagnostic, predictive and prescriptive analysis in order of increasing complexity.
What an analyst does — and the four-type ladder
The analyst's job: support decisions with data
A data analyst's real job is not to build pretty charts for the sake of it. The core purpose is to help other people make better, evidence-based decisions. Without an analyst, decisions get made on gut feel or habit. With one, they are grounded in facts. But here is a critical point: the analyst prepares the decision — they do not make it instead of the manager. The manager owns the outcome; the analyst owns the evidence.
There are four types of analysis, arranged in a ladder of increasing complexity and increasing need for human judgement:
1. Descriptive analysis answers "What happened?" — for example, "Coffee-shop sales in March fell 20% compared to February." This is the most basic level: it simply states a fact from the data.
2. Diagnostic analysis answers "Why did it happen?" — for example, "Sales fell because a competitor opened next door and our lead barista was on holiday." Here we dig for causes behind the fact.
3. Predictive analysis answers "What will likely happen?" — for example, "If the competitor stays, April sales may fall a further 10–15%." This is a probabilistic forecast, not a guarantee.
4. Prescriptive analysis answers "What should we do?" — for example, "We recommend launching a loyalty programme and hiring a second barista." This is the most complex level: it requires data, business context, and judgement about goals and constraints.
As you move up the ladder, the analysis gets more powerful — and more demanding. Most real-world analytics work starts at the bottom and builds upward.
Lesson notes
The analyst's job: support decisions with data
A data analyst's real job is not to build pretty charts for the sake of it. The core purpose is to help other people make better, evidence-based decisions. Without an analyst, decisions get made on gut feel or habit. With one, they are grounded in facts. But here is a critical point: the analyst prepares the decision — they do not make it instead of the manager. The manager owns the outcome; the analyst owns the evidence.
There are four types of analysis, arranged in a ladder of increasing complexity and increasing need for human judgement:
1. Descriptive analysis answers "What happened?" — for example, "Coffee-shop sales in March fell 20% compared to February." This is the most basic level: it simply states a fact from the data.
2. Diagnostic analysis answers "Why did it happen?" — for example, "Sales fell because a competitor opened next door and our lead barista was on holiday." Here we dig for causes behind the fact.
3. Predictive analysis answers "What will likely happen?" — for example, "If the competitor stays, April sales may fall a further 10–15%." This is a probabilistic forecast, not a guarantee.
4. Prescriptive analysis answers "What should we do?" — for example, "We recommend launching a loyalty programme and hiring a second barista." This is the most complex level: it requires data, business context, and judgement about goals and constraints.
As you move up the ladder, the analysis gets more powerful — and more demanding. Most real-world analytics work starts at the bottom and builds upward.