Data Mining BABOK Technique 10.14: Elements, BA Role, Exam Angle

Data Mining is technique 10.14 in the Business Analysis Body of Knowledge (BABOK Guide v3). It uses analytical and statistical methods to find useful patterns in large data sets so that people make better decisions. In data mining BABOK terms, the work has five elements: requirements elicitation, data preparation, data analysis, modelling techniques and deployment. The business analyst (BA) frames the goal and lands the result; specialists usually build the model.

Last verified: October 2026 against the IIBA BABOK Guide page, the BABOK v3 errata and the CBAP exam page [1, 2, 3].

What the data mining technique is in BABOK v3

Data mining in BABOK v3 is an analytic process that looks at large amounts of data from different angles and summarises it into patterns and relationships that support a decision (BABOK 10.14.1 Purpose, 10.14.2 Description).

The guide treats it as a decision-support technique, not as a data-science method in its own right. The output is meant to feed something the business uses: a decision model, a set of business rules, a forecast, or a better understanding of current performance. That is why the technique sits close to decision modeling and decision analysis in practice: a mined pattern often becomes a rule inside a decision.

The description also says a mining effort can be run top-down (testing a specific idea), bottom-up (exploring with no fixed hypothesis), or as a mix of both (10.14.2).

BABOK groups the results into three kinds of analysis. Descriptive analysis shows what the patterns are, for example through clustering. Diagnostic analysis explains why a pattern exists, for example with decision trees. Predictive analysis estimates how likely something is in the future, for example with regression or neural networks (10.14.2). The differences between the three are a topic of their own; here they matter only as a way to name the goal of a mining effort.

Why data mining matters for the CBAP exam

Data mining is one of the 50 BABOK techniques, and CBAP questions test whether you know when it fits and who does what, not whether you can run an algorithm.

The Certified Business Analysis Professional (CBAP) exam has 120 case-study and scenario-based multiple-choice questions in 210 minutes [3]. Techniques are not weighted separately. They appear inside scenarios tied to knowledge-area tasks. Data mining is listed under tasks in more than one knowledge area; the map of BABOK techniques by knowledge area shows every task where it appears.

Typical exam signals for data mining:

• a large volume of transactional or historical data already exists;
• stakeholders need patterns or predictions, not opinions;
• the scenario asks what the BA should do first (answer: clarify the decision and the goal), or what limits the result (answer: data quality, access, explainability).

Data mining BABOK elements: how the technique works

The data mining technique runs in five elements, from setting the goal to putting the model into use (10.14.3).

.1 Requirements elicitation

The mining effort starts with a goal. BABOK frames that goal and its scope either as decision requirements for a specific business decision or around a functional area (10.14.3 .1) [8]. Good questions at this point: which decision will change, who makes it, how often, and what result would be good enough to act on.

.2 Data preparation

Data from several tables or systems is merged into one wide analytical data set [8]. Part of the data is usually held back so the model can be checked on records it has not seen (10.14.3 .2). The BA's contribution here is meaning: what each field represents, which values are valid, and which sources are trustworthy. A data dictionary and a deliberate approach to sourcing and validating data make this step faster and safer.

.3 Data analysis

Analysts explore the data to find characteristics and variables that carry signal. A practitioner guide written by a business consultant calls this step "often the longest and most complex" [8]. Survey data agrees on where the time goes: in a 2020 survey of nearly 2,400 data professionals, data cleansing took more than a quarter of an average day and data loading 19%, while model selection, training and deployment each took 11–12% [5].

.4 Modelling techniques

BABOK names a set of techniques at overview level, split into supervised and unsupervised approaches (10.14.3 .4). The ones a CBAP candidate should recognise:

Teams often try several techniques and keep the one that performs best on the held-back data [8]. One detail for careful readers: the IIBA errata corrects 10.14 .4, where the printed guide says "support sector machines" instead of support vector machines [2].

.5 Deployment

A model has value only when it is used. BABOK describes deployment as putting the model or its rules into business operations, for example as rules in a decision, a score in a system or a regular report (10.14.3 .5). This is where most of the BA's classic work comes back: requirements for the integration, transition, training and measuring whether decisions really improved.

Usage considerations: strengths and limitations

BABOK presents data mining as strong at revealing hidden patterns and weak when people apply it without understanding it (10.14.4).

Two limitations deserve extra attention on the exam and at work. Patterns are not causes; and data that survived some filter can mislead, as the survivorship bias example from churn surveys shows. Success also depends on having "the right type, volume and quality of data" [8].

What happens in reality: most models never ship

In real projects the weakest step is usually deployment, not modelling. In the 2023 Rexer Analytics survey, only 22% of data scientists said their new, "revolutionary" initiatives usually deploy, and only 32% said their models in general usually deploy [6]. The author who reported the survey links the failures to "endemic under-planning and business stakeholders lacking concrete visibility" [6]. Only 49% of respondents said stakeholders have enough knowledge most of the time or always [6].

A 2021 column in a data science journal makes the same point from another angle: projects that start with the algorithm are "where many deployments fail," and data science training covers stakeholder analysis and change management poorly [7]. Its example is a large routing project where most of a ten-year implementation and its several-hundred-million-dollar budget went into deployment and change management [7].

A widely used industry process for mining work, CRISP-DM (Cross-Industry Standard Process for Data Mining, published in 1999), has six phases from business understanding to deployment [4]. It led every methodology poll on a major data-mining site from 2002 to 2014, yet critics call it rigid, documentation-heavy and weak on stakeholder communication [4]. Its phases line up closely with the five BABOK elements, which makes it easy to map the BA's role onto a data team's own process.

Experts disagree on how far the BA goes into the data

Practitioners agree that the BA owns the goal and the deployment, but not on how much of the analysis the BA should do personally.

1. The BA frames the decision and lands the change. The data team builds models; the BA defines the decision, the success measure and the adoption plan. Survey and journal authors who study deployment failure take this line, because the gaps they find are planning and stakeholder gaps, not algorithm gaps [6, 7].

2. The BA should do light analysis directly. A supply-chain analytics lead argues that predictive work "isn't a departure from business analysis" and that BAs can use spreadsheets and BI tools without becoming full-time data scientists [10]. A BA training provider goes further and describes predictive and prescriptive analytics as part of the expanded BA role [11].

3. Query skills, yes; building data assets, no. A CBAP-certified consultant with a background in financial institutions holds that SQL "can greatly increase your effectiveness as a BA", but leaves building databases to technical roles and says the answer depends on how the enterprise is organised [9].

The real disagreement is between positions 1 and 2: whether a BA who builds a regression in a spreadsheet adds value or takes on a specialist's risk. BABOK's own limitation, that using techniques without understanding them produces false correlations, supports caution (10.14.4).

Practice Questions

Question 1

A retailer wants to use three years of loyalty-card data to decide which customers receive a retention offer. The sponsor asks the business analyst to start a data mining effort. What should the BA do first?

A. Ask the data team to run clustering on all available data B. Define the decision to be supported, its scope and what result would be good enough C. Merge all customer tables into one analytical data set D. Select a neural network because it handles complex patterns

Answer: B. The first element, requirements elicitation, sets the goal and scope, often as decision requirements (10.14.3 .1). A and D jump to modelling; C is data preparation and comes later.

Question 2

A model predicts loan default accurately on test data, but loan officers ignore its scores because they cannot see why a case is flagged. Which BABOK limitation does this illustrate?

A. Data volume was too small B. Some techniques are hard to explain, so stakeholders may not trust the results C. Data preparation was skipped D. The wrong exam technique was chosen

Answer: B. BABOK lists stakeholder trust in results that come from complex methods as a limitation (10.14.4). A decision tree or added explanation could help adoption.

Question 3

A team found that customers who use the mobile app churn less. A stakeholder proposes forcing all customers onto the app to cut churn. What is the BA's best response?

A. Approve the proposal because the pattern is strong B. Point out that the pattern shows an association, not a cause, and propose a test C. Ask for a larger data set and repeat the same analysis D. Replace the model with a regression

Answer: B. Mined patterns can be misread as causes (10.14.4). The BA should treat the pattern as a hypothesis to test before changing the process.

Conclusion: the gap between BABOK data mining and real projects

BABOK describes a complete cycle from goal to deployment; real projects spend most of their time on data and too little on the two ends where the BA leads.

For the exam, the BA in a data mining scenario defines the decision, protects data meaning and quality, questions patterns before acting on them, and plans adoption. In practice, those same tasks are the ones data teams most often leave out.

Related topics

• CBDA exam guide: format, domains and difficulty, the IIBA certification that goes deeper into analytics work.
• The CBAP exam simulator and overview and the CBDA exam page for practice questions.

FAQ

What is data mining in BABOK?

Data mining is BABOK v3 technique 10.14. It uses analytical and statistical methods to find patterns in large data sets that support business decisions. Its five elements are requirements elicitation, data preparation, data analysis, modelling techniques and deployment.

Does a business analyst need to build data mining models?

No. BABOK describes the technique, not who runs the algorithm. In most teams data scientists build models, while the BA defines the decision, clarifies data meaning, sets acceptance criteria and plans deployment. Some employers expect BAs to do light analysis in SQL or BI tools as well.

Which modelling techniques does BABOK mention for data mining?

At overview level, BABOK covers supervised and unsupervised techniques such as classification, regression, clustering, association rules, decision trees, neural networks and support vector machines. The CBAP exam expects you to recognise what each answers, not to compute them.

What are the main limitations of data mining?

Applying techniques without understanding them can produce false correlations. Data, tools and specialists can be expensive. Complex models can be hard to explain, which makes stakeholders slow to trust and use them.

How is data mining tested on the CBAP exam?

It appears inside scenarios, not as a definition test. Expect questions on when data mining fits, what the BA should do first, and which limitation explains a problem. The CBAP exam has 120 scenario-based questions in 210 minutes.

Sources

1. BABOK Guide
2. BABOK Guide Errata
3. Certified Business Analysis Professional (CBAP)
4. What is CRISP DM?
5. Data Prep Still Dominates Data Scientists' Time, Survey Finds
6. Survey: Machine Learning Projects Still Routinely Fail to Deploy
7. Column Editors' Note on model deployment
8. BA Techniques: Data Mining
9. Do Business Analysts Need to Know SQL?
10. Predictive Analytics for Business Analysts: How to Lead with Insight, Not Assumptions
11. Harnessing the Power of Data Analytics in Business Analysis