DDatalytIQsCA35P Topic Workspace
Menu
Sign in
← CA35P dashboardTOPIC 02 · FREE ACCESS

Introduction to Data Analytics

Analytics foundations, data lifecycle, big data, tools and visual communication.

GUIDED LEARNING · APPROX. 210 MINUTES

Learn, practise and produce evidence

Each lesson combines examinable concepts, a worked example, an applied exercise and an answer guide.

PRACTICE FILES

Download the controlled source files

Open the CSV files in Excel, then save your working version as an XLSX workbook.

CRISP engagement briefCSV · decision, data and lifecycle fieldsDownload ↓Visualisation practice dataCSV · monthly channel performanceDownload ↓
2.1

CRISP framework and governed data lifecycle

60 minutes
not started

By the end, you should be able to:

  • Translate a management question into CRISP phases and analytical tasks.
  • Distinguish conceptual, logical and physical data models.
  • Specify controls across acquisition, use, retention and disposal.

CRISP discipline

Business understanding anchors the work; data understanding, preparation, modelling, evaluation and deployment must remain traceable to the decision.

Model layers

Conceptual models define entities, logical models define relationships and rules, and physical models define implementation.

Lifecycle governance

Ownership, lawful purpose, quality, access, retention and disposal controls apply from sourcing to deletion.

WORKED EXAMPLE

A lender wants to reduce arrears without excluding viable customers.

  1. Define the decision and measurable success criterion.
  2. Inventory customer, repayment and interaction data.
  3. Record quality, privacy and representativeness risks.
  4. Specify evaluation and deployment monitoring.
Decision → CRISP phase → evidence → control → accountable owner

Interpretation: A technically accurate model is not decision-ready unless its purpose, controls and deployment consequences are explicit.

DATASET MISSION · CRISP engagement brief

Complete the decision, success_measure, data_sources, quality_risks and phase_deliverables rows for a school fee-arrears support project. Specify a non-punitive action and an accountable owner for each phase.

Reveal source-based answer guide

A defensible decision is which families should receive optional payment support. Define an outcome and time horizon; use only authorised billing data; test missing records and uneven treatment; assign school finance/data-protection owners and evaluate support outcomes before deployment.

Further practice

Complete the CRISP engagement brief for a school seeking to improve fee collection while protecting vulnerable learners.

A strong response defines the decision, identifies payment and learner-context data, addresses missingness and fairness, selects descriptive and predictive outputs, and assigns monitoring controls.

Sign in to record progress
2.2

Big data and analytical decision types

45 minutes
not started

By the end, you should be able to:

  • Assess a dataset against the five Vs.
  • Choose descriptive, predictive or prescriptive analysis for a stated decision.

Five Vs

Volume, velocity, variety, veracity and value describe scale and management difficulty rather than quality by themselves.

Descriptive

Explains what happened through summaries, patterns and exceptions.

Predictive and prescriptive

Prediction estimates likely outcomes; prescription evaluates actions subject to constraints.

WORKED EXAMPLE

A retailer receives transactions, web events and customer-service messages continuously.

  1. Classify volume, velocity and variety.
  2. Test veracity using completeness and consistency checks.
  3. Describe current abandonment patterns.
  4. Predict high-risk sessions and evaluate interventions.
Value = decision benefit − data, model and control cost

Interpretation: More data is useful only when it improves a defined decision sufficiently to justify its risks and costs.

DATASET MISSION · Visualisation practice data

Calculate conversion rate for Organic and Paid in January and April. Name the five Vs implicated by an added live clickstream with duplicate event IDs, and choose the first descriptive analysis to publish.

Reveal source-based answer guide

January Organic 744/12,400 = 6.0%, Paid 490/9,800 = 5.0%. April Organic 704/15,300 ≈ 4.60%, Paid 500/13,900 ≈ 3.60%. Duplicates threaten veracity; live arrival is velocity; scale is volume. Publish rates with denominators before predicting.

Further practice

Classify three proposed analyses as descriptive, predictive or prescriptive and justify the input data each needs.

Past-sales dashboard is descriptive; default-risk score is predictive; constrained inventory allocation is prescriptive.

Sign in to record progress
2.3

Analytics technology and architecture

45 minutes
not started

By the end, you should be able to:

  • Match tools to cleaning, storage, analysis and reporting tasks.
  • Explain when spreadsheet, database, cloud or specialist tooling is proportionate.

Data preparation

Excel or Power Query can support controlled moderate-volume cleaning; scripted pipelines improve repeatability at scale.

Storage

Relational databases enforce structure and integrity; object storage supports diverse large files.

Reporting

BI platforms distribute governed metrics, while notebooks support exploratory and reproducible analysis.

WORKED EXAMPLE

An NGO consolidates 30 monthly workbooks from county teams.

  1. Standardise a submission schema.
  2. Use Power Query for repeatable ingestion.
  3. Store validated records in a relational table.
  4. Publish governed indicators to a dashboard.
Tool fit = data scale + repeatability + governance + user capability

Interpretation: The best tool is the least complex option that reliably satisfies scale, auditability and delivery requirements.

DATASET MISSION · Visualisation practice data

In Excel, create a tidy month-channel table with calculated conversion_rate. Independently reconcile each month’s total conversions and revenue to the chart source. Document refresh date and source owner.

Reveal source-based answer guide

January conversions 1,234 and revenue KES 3,702,000; April conversions 1,204 and revenue KES 3,612,000. Keep a refresh control and avoid mixing conversion counts with monetary values on one unlabelled axis.

Further practice

Design a tool chain for a monthly programme dataset with 50,000 rows, recurring corrections and executive reporting.

A defensible design uses controlled templates, automated ingestion/validation, database storage and a governed BI dashboard.

Sign in to record progress
2.4

Decision-focused data visualisation in Excel

60 minutes
not started

By the end, you should be able to:

  • Select charts for comparison, trend, composition and relationship.
  • Remove misleading scales, clutter and unsupported precision.
  • Write a decision-led chart title and annotation.

Chart purpose

Bars compare categories, lines show ordered time, scatter plots show relationships and limited-part pie or stacked charts show composition.

Integrity

Axes, units, baselines, filters and missing values must be visible and consistent.

Narrative

Titles state the finding; annotations explain material changes without overstating causality.

WORKED EXAMPLE

Monthly conversion falls despite increasing website traffic.

  1. Create a line chart for traffic and conversion.
  2. Use a secondary axis only if clearly labelled and necessary.
  3. Annotate the month the process changed.
  4. State association rather than unsupported causation.
Signal-to-noise = decision-relevant ink ÷ total visual ink

Interpretation: A chart succeeds when a decision-maker can identify the pattern, scale and qualification without verbal rescue.

DATASET MISSION · Visualisation practice data

Design two charts: monthly visits by channel and conversion rate by channel. Explain why a rise in Organic visits from January to April does not establish better conversion performance.

Reveal source-based answer guide

Organic visits increase 12,400 → 15,300 while conversions fall 744 → 704; conversion rate declines from 6.0% to about 4.60%. Label denominators and rates, and do not infer cause from four months of observations.

Further practice

Use the supplied data to produce one comparison chart and one trend chart, then write a two-sentence management interpretation.

The answer should use correct chart types, labelled units, consistent periods, restrained formatting and a conclusion tied to the observed values.

Sign in to record progress
CASE QUESTION BANK

Apply the method before the quiz

4 original case questions with answer guides. Self-check practice does not change your assessment record.

  1. A credit team asks for a default prediction but has no definition of default. Which CRISP-DM phase is incomplete?

    Show worked answer

    Business understanding. Define the outcome, time horizon, cost of errors and decision before selecting data or a model.

  2. Describe the difference between a logical and physical model for customer payments.

    Show worked answer

    A logical model defines entities, keys and relationships independent of technology; a physical model specifies tables, data types, indexes and implementation constraints.

  3. A streaming dataset has millions of fast-arriving records but many duplicate IDs. Which of the five Vs raises the primary trust concern?

    Show worked answer

    Veracity. Assess duplication, completeness and validity; volume and velocity do not establish reliability.

  4. Which analysis supports choosing a constrained allocation of limited audit hours?

    Show worked answer

    Prescriptive analytics compares feasible actions under capacity and risk constraints; descriptive summaries and predictions can inform its inputs.

DATASET SELF CHECK

Calculate, enter and interpret

Use the downloadable CSV, then enter your result. These checks give immediate feedback and do not affect your scored quiz or learner record.

Visualisation practice data

January Organic conversion rate (percentage points)?

Visualisation practice data

April Organic conversion rate (round to two decimals)?

AUTO-SCORED ASSESSMENT

Topic 2 knowledge check

Five scored questions · pass mark 80% · unlimited attempts. Your latest score is retained with the attempt history.

Sign in to take the quiz
ASSESSED PRACTICAL

Submit an analytics engagement and visualisation brief

Use the CRISP template and visualisation dataset to define the decision, lifecycle controls, tool chain and two management-ready charts.

Assignment deliverables

  • CRISP-DM decision brief with success measure and phase deliverables
  • Conceptual, logical and physical model sketches with lifecycle controls
  • Two labelled charts and a justified descriptive, predictive or prescriptive choice

Use the downloadable practice data. Check source totals, state assumptions, label figures and remove personal data before sharing evidence.

20-mark rubric

Structure, controls and data integrity 5Analytical method and accuracy 5Interpretation of evidence 5Decision recommendation and presentation 5
Sign in to submit evidence