Financial Management and Business Data Analytics · Introduction to Data Science for Business Decision-making

Descriptive, Predictive and Prescriptive Analytics Explained

Updated 10 October 2026 · Fact-checked

Descriptive analytics summarises what happened. Predictive analytics uses past data and models to estimate what is likely to happen. Prescriptive analytics recommends the best action given goals and constraints. To solve a question, identify the business question being asked, then match it to the type and a suitable technique.

Understand Analytics Techniques: Descriptive, Predictive, Prescriptive

Business analytics means using data to support decisions. It is usually grouped by the question it answers. Think of a ladder: first you see the past, then you look ahead, then you choose an action.

Descriptive analytics answers "What happened?" It uses totals, averages, percentages, charts and dashboards. Example: monthly sales by region for a Pune retailer, or the average days customers take to pay. It gives no forecast and no advice.

Predictive analytics answers "What is likely to happen?" It finds patterns in historical data and uses them to estimate future or unknown values. Techniques include regression, time series forecasting and classification. Example: estimating next quarter's sales, or the chance that a borrower will default. The output is an estimate, not a certainty.

Prescriptive analytics answers "What should we do?" It combines predictions with objectives and constraints to recommend an action. Techniques include optimisation (for example linear programming), simulation and decision rules. Example: choosing the product mix that maximises profit within machine-hour limits.

Some syllabus material also names diagnostic analytics, which answers "Why did it happen?" using drill-down and correlation. Treat it as a step between descriptive and predictive.

Machine learning is a set of methods where a model learns patterns from data instead of being given fixed rules. In supervised learning the data has known outcomes (labels), for example past loans marked as repaid or defaulted; it is used for regression (predicting a number) and classification (predicting a category). In unsupervised learning there are no labels, and the model finds structure, for example clustering customers into segments. Models should be tested on data they have not seen, to check they are not just memorising the past (overfitting).

Key rules to remember

Descriptive analytics: Question: What happened? Tools: mean, median, totals, ratios, charts, dashboards; Looks backward. No forecast or recommendation.
Predictive analytics: Question: What is likely to happen? Tools: regression, time series, classification; Uses historical data. Output is a probability or estimate, not a guarantee.
Prescriptive analytics: Question: What should we do? Tools: optimisation, simulation, decision rules; Needs an objective and constraints. Builds on predictions.
Simple linear regression: Y = a + bX; Predicts a numeric value Y from X. Regression shows association; it does not by itself prove cause.
Supervised vs unsupervised learning: Supervised: labelled outcomes. Unsupervised: no labels; Regression and classification are supervised. Clustering is unsupervised.

How to solve Analytics Techniques: Descriptive, Predictive, Prescriptive questions

Use this method for any question that asks you to classify, explain or apply an analytics type.

1Read the scenario and underline the business question in your own words.
2Decide the time focus: past (what happened), future (what will happen) or action (what to do).
3Name the type: descriptive, predictive or prescriptive. Mention diagnostic if the question is about why.
4Choose a technique that fits: summary statistics or charts, regression or forecasting, or optimisation or simulation.
5State the data needed and the output you would get, using the scenario's figures or names.
6Say how the result helps the decision, and note one limit, such as data quality or forecast error.
7Close with a one-line conclusion linking type, technique and decision.

Quickest way: Question-word shortcut

When to use it: Use for MCQs and for the first line of a written answer.

Look for the key verb: summarise, report, show = descriptive.
Estimate, forecast, predict, likelihood, classify = predictive.
Recommend, optimise, best, allocate, maximise within limits = prescriptive.
Why, root cause, drill down = diagnostic.
Known outcomes in the data = supervised; grouping with no outcomes = unsupervised.

Common mistakes in Analytics Techniques: Descriptive, Predictive, Prescriptive

Calling a sales forecast descriptive analytics
Forecasts use past data, so students think they only describe the past.
Fix: Ask whether the output is about the future. If yes, it is predictive.
Treating predictive output as a certain result
Numbers look exact.
Fix: Say it is an estimate with error, based on past patterns that may change.
Saying prescriptive analytics only predicts
Students confuse it with predictive analytics.
Fix: Prescriptive must end in a recommended action, using objectives and constraints.
Mixing up supervised and unsupervised learning
Both names sound similar.
Fix: Supervised has labelled outcomes to learn from. Unsupervised has none and finds groups or patterns.
Writing definitions without an example
Students memorise text and skip application.
Fix: Add one Indian business example for each type, using the scenario given.
Claiming correlation or regression proves cause
A strong fit feels like an explanation.
Fix: Say it shows association and that cause needs further reasoning or testing.

Worked examples

Example 1

A Chennai retail chain does the following: (a) prepares a report of last year's sales by store; (b) estimates next quarter's sales using past data and festival timing; (c) decides how many units of each product to send to each store to minimise transport cost within warehouse limits. Classify each and name a technique.

Show the solution

(a) The question is what happened. This is descriptive analytics. Technique: totals, averages and a bar chart by store.
(b) The question is what is likely to happen. This is predictive analytics. Technique: time series forecasting or regression on past sales.
(c) The question is what to do under constraints. This is prescriptive analytics. Technique: optimisation such as a transportation model using linear programming.
Link: the forecast in (b) can feed the demand figures in (c).

Answer: (a) Descriptive, summary statistics and charts. (b) Predictive, forecasting or regression. (c) Prescriptive, optimisation. Each answers a different question: what happened, what will happen, what to do.

Example 2

A bank has data on 10,000 past loans, each marked as repaid or defaulted. It builds a model to estimate whether a new applicant will default. Identify the analytics type and the learning approach, and state one precaution.

Show the solution

The bank wants to estimate a future outcome for a new applicant. This is predictive analytics.
The past loans carry known outcomes (repaid or defaulted), so these are labels. This is supervised learning.
The result is a category (default or not), so the task is classification, not numeric regression.
Precaution: test the model on loans not used to build it, so you know it works on new cases and has not just memorised old data.
Further point: the bank can later use the predicted risk with its rules to decide approval or interest rate. That step would be prescriptive.

Answer: Predictive analytics using supervised learning (classification). Test it on unseen data to avoid overfitting. Using the result to set loan decisions would be prescriptive.

Exam tips

MCQs usually give a short scenario. Find the question word first, then choose the type.
In written answers, define each type in one line, then give an example from the scenario. This earns step marks.
Learn one technique for each type: summary statistics, regression or forecasting, and optimisation.
Be ready to separate supervised and unsupervised learning with one example each.
Mention limits such as data quality, bias and forecast error in one line to show judgement.

Practice questions from Introduction to Data Science for Business Decision-making

Analytics Techniques: Descriptive, Predictive, Prescriptive in other exams

The same ground in other exams, if you are preparing for more than one or want another angle on it.

Analytics Techniques: Descriptive, Predictive, Prescriptive: frequently asked questions

What is the difference between descriptive and predictive analytics?

Descriptive analytics summarises past data to show what happened. Predictive analytics uses that past data in a model to estimate what is likely to happen next. Descriptive gives facts; predictive gives estimates.

Is prescriptive analytics the same as predictive analytics?

No. Predictive analytics estimates outcomes. Prescriptive analytics goes further and recommends the best action given goals and constraints. It often uses predictions as an input.

What are machine learning basics I need for CMA Intermediate?

Know that machine learning lets a model learn patterns from data. Know supervised learning (labelled data, used for regression and classification) and unsupervised learning (no labels, used for clustering). Also know the need to test on unseen data.

Can you give simple examples of each type of analytics?

A monthly sales dashboard is descriptive. A model that forecasts next quarter's demand is predictive. A model that picks the profit-maximising product mix within resource limits is prescriptive.