CMCE10002 Foundations of Business Analytics
The Complete Exam Bible · S2 2026CMCE10002 Overview
- The University of Melbourne
- Semester 2 2026
- First-year commerce subject
- Taught in R and Posit Cloud
- 10 hours per week expected
Foundations of Business Analytics at the University of Melbourne
CMCE10002 Foundations of Business Analytics is a first-year subject in which students build the skills to read data and draw business decisions out of it.
- Assessed by in-class quizzes, a mid semester test, a group assignment and a two-hour hand-written final examination
- The one condition to plan around the subject guide requires a pass in the final exam, so a strong semester mark does not cover a failed paper
- Hardest step predicting what a table looks like after grouping, reshaping or joining, without a computer in front of you
- How to prepare cover the output, predict it, then run the code; both timed assessments are sat without a console
How CMCE10002 is assessed
| Component | Weight | Format |
|---|---|---|
| In-class quizzes, held during workshops | 10% | Individual, weeks 4 to 12. Each week is worth one mark, half for attending your workshop and half for contributing, with randomised preparation spot checks. Nine quizzes run and the best seven count. |
| Mid Semester Test | 20% | Individual, week 6. One hour with no reading time, sat in person as a Canvas quiz under the Respondus LockDown browser, 25 questions mixing multiple choice and short answer, covering the first five weeks. |
| Group Assignment | 20% | Group, week 12. An analytics project on a business dataset, reported as an Analytics Team Memo, written for a peer or manager who will check the work, and a short Corporate Brief, written for an executive who has to decide something. |
| 2-hour end of semester examination · hurdle | 50% | Individual, during the exam period. Hand written, drawing on lectures, workshops and the assigned readings: 20 multiple-choice questions worth 40 per cent, alongside roughly four short-answer questions that carry subparts and the remaining 60 per cent. The subject guide states that a pass in this exam is required to complete the subject. |
The four weights published in the subject guide for Semester 2, 2026 sum to 100 per cent. The pass condition printed with the final examination row is the only one stated for any component; no other component is described as carrying one, and the subject site holds the current wording. Two official statements about the group differ: the subject guide describes groups of four while the week 6 lecture slides describe groups of three to four registered in the same class, so confirm the current rule before your group is finalised.
Assessment structure
Current CMCE10002 dates
| Date | Item | Control |
|---|---|---|
| Friday 4 September 2026, 17:30 to 18:30 AEST | Mid semester test | One hour with no reading time, sat in person as a Canvas quiz under the Respondus LockDown browser. |
| 5 pm each Friday, weeks 4 to 12 | Weekly quiz due | Nine weekly quizzes run in total and the best seven contribute to the final grade. |
| Monday of week 9 | Group assignment released | Groups are finalised by the Friday of week 9. |
| 23:59 on the Friday of week 12 | Group assignment due | Late submission without an approved extension or special consideration receives zero marks. |
| During the examination period | Final examination | The subject guide places the exam in the examination period; confirm your own sitting on the University timetable. |
Confirm current dates and operational instructions on the live subject site.
What CMCE10002 covers
Foundations of Business Analytics teaches analytics as a chain of decisions: a business question becomes an analytical plan, the plan is implemented in R, and the output is read back as something a manager can act on.
The chapters follow the twelve lecture weeks, from the three kinds of analytics question through visualisation, wrangling, reshaping and joining, to causal and predictive work and the two written products of the group assignment.
What Business Analytics Is: Three Kinds of Question
Lecture week 1. Descriptive, predictive and causal questions; reading the verb in a request; the evidence each type demands; the four layers from raw record to a narrative; bias, rights and transparency as constraints.Three dimensions, three different evidence bars
Foundations of Business Analytics opens by splitting the field into three dimensions: descriptive analytics, which establishes what happened in the past; predictive analytics, which anticipates what happens next; and causal analytics, which uncovers the cause-and-effect relationships that explain why something happened.
- Descriptive, predictive and causal questions, and how to tell them apart
- Reading the verb in a request to identify the question type
- The evidence each question type demands before it can be answered
- The four layers from raw record to a narrative a manager can act on
- Bias, individual rights and transparency as constraints on the method
Reproducible Workflows in Posit Cloud and Quarto
Lecture week 2. Posit Cloud as the shared environment; console, script and Quarto document; loading packages; rendering from a clean session; reading the first error rather than the last.Everything runs in one shared environment
The subject uses R and RStudio through Posit Cloud, a browser-based environment that needs no installation, saves work online and is reachable from any device.
- Posit Cloud as the shared environment for every exercise and assignment
- Console, script and Quarto document, and what each one preserves
- Loading packages so their functions exist for the session
- Rendering from a clean session as the test of reproducibility
- Reading the first error rather than the last one
Reading R Code: Objects, Tibbles and Subsetting
Lecture week 2 and workshop week 3. Object, vector, tibble, package and argument; the symbol set and the sentence each makes; row and column positions in subsetting; establishing the grain before computing.Five words that make every later line readable
The class quick reference opens with five ideas: an object is anything held under a name, a vector holds values of one type, a data frame or tibble is a table of rows and columns, a package bundles extra functions, and an argument is an input handed to a function.
- Object, vector, tibble, package and argument as the working vocabulary
- The symbol set, and the sentence each symbol makes
- Addressing rows and columns inside square brackets, and the silent blank
- The small set of functions worth knowing without hesitation
- Establishing the grain of a dataset before computing anything from it
Visualising Business Data with ggplot2
Lecture week 3. The retail pricing question the demonstration is built on; data, mapping, geometry and polish; why the canvas starts empty; mapping a column to colour against setting a constant; matching a geometry to the question.A plot is built in four parts
Week 3 opens with a business challenge, not a chart.
- The business question the week 3 demonstration is built around
- Data, mapping, geometry and polish as four separable parts
- Why naming the data alone renders an empty plot area
- Mapping a column to colour, against setting a colour as a constant
- Matching a geometry to the question rather than to the data type
Why Summary Statistics Alone Can Mislead
Lecture week 3. Datasets that share a mean and a spread while differing completely; what an average can and cannot establish; reading a table of group averages; separating a spurious pattern from a hidden spread.The summary table agreed and the data did not
Week 3 does not open with a chart function.
- Why the week 3 lecture puts visualisation before analysis
- Datasets that share a mean and a standard deviation while differing completely
- What an average and a spread can and cannot establish
- Reading a table of group averages without overreaching
- Distinguishing a spurious pattern from a hidden spread
Wrangling One Table with dplyr
Lecture week 4. Wrangling as transformation fit for one purpose; row, column and group operations; filtering before ranking; turning a dollar total into a comparable rate; selecting by name, pattern and type.Transformation with a stated purpose
The week 4 lecture defines wrangling as reshaping a messy extract on purpose, until the table is clean, well structured and fit for one specific purpose.
- Wrangling defined as transformation fit for one stated purpose
- Row, column and group operations as three pairs of verbs
- Filtering before ranking, and why the order is not stylistic
- Turning a dollar total into a comparable rate
- Selecting by name, by pattern and by type, and which survives a change in the data
The Pipe and Grouped Business Summaries
Lecture week 4. The pipe against nested calls; grouping as a label that changes nothing visible; summarising as the step that changes the grain; carrying a count beside every average; releasing the grouping afterwards.Read the pipe as the word then
Real wrangling is rarely one verb.
- The pipe as a readable alternative to nested function calls
- Grouping as a label that leaves the table looking untouched
- Summarising as the step that changes the grain of the table
- Carrying a count beside every average, and why it is not optional
- Releasing the grouping so later steps do not run per group
Tidy Data and Reshaping with Pivots
Lecture week 5. The three tidy rules; why tidy data lets code survive more data; lengthening when a variable sits in a column heading; widening when a variable name sits in a cell; predicting the reshaped row count.Tidy is a contract, not tidiness
A dataset is tidy when each column is a variable, each row is an observation and each cell is one value.
- The three tidy rules and what each one rules out
- Why tidy data makes code survive the arrival of more data
- Lengthening a table when a variable sits in the column headings
- Widening a table when a variable name is stored as data
- Predicting the row count of a reshaped table before running it
Joining Tables on a Shared Key
Lecture week 5. The key as a statement about what a row represents; matched, left-only and right-only keys; what each of the four joins keeps; why order matters for two of them; matching grain before joining.The join type is a decision about unmatched keys
Two tables can be combined when they share a column, or set of columns, identifying what a row represents.
- The key as a statement about what a row represents
- Matched keys, left-only keys and right-only keys as three regions
- What each of the four joins retains, and which columns travel with it
- Why order matters for two of the joins and not the other two
- Matching grain before the join, and predicting the row count after it
Variation in Cross Section, Time Series and Panel Data
Lecture week 6. Three readings of one table; what one observation is in each; grouping by the unit against grouping by the period; why a table with one column per period must be lengthened first.One table, three ways of using it
The week 6 demonstration puts up a single grid of units against periods and reads it three ways. A cross section compares units at one moment, a time series follows one unit across moments, and a panel has both.
- Cross section, time series and panel as three readings of one table
- What one observation is in each of the three shapes
- Grouping by the unit, and the variation that survives
- Grouping by the period, and the variation that survives
- Why a table with one column per period must be lengthened first
Causal Analytics and Randomised Experiments
Lecture week 8. Spotting a causal question; potential outcomes and the one that is never observed; omitted variables; model-based adjustment against randomised design; comparability and stability.The comparison you need is with a world that did not happen
A causal question asks whether one thing produced another, and the subject gives a reliable tell: the analysis is being used to argue for changing a procedure, policy or practice.
- Spotting a causal question from the decision it is meant to support
- Potential outcomes, and why only one of them is ever observed
- Omitted variables that drive both the treatment and the outcome
- Model-based adjustment against design-based randomisation
- Comparability and stability, and telling which one breaks in a given setting
Predictive Analytics, Stored Data and APIs
Lecture weeks 9 to 11. Forming a rule and evaluating it as separate jobs; holding back later rows; comparing against a trivial alternative; describing the slice taken from a store; keeping a live data request reproducible.Three topics, one shared job
The last three teaching weeks name using data to form and evaluate forecasts, storing and using existing data, and collecting structured and unstructured data from the interfaces services publish.
- Forming a rule and evaluating it as two separate jobs
- Holding back later rows when the data runs in time
- Comparing every forecast against a trivial alternative
- Describing the slice taken from a shared store precisely
- Recording the request and saving the response so live data stays reproducible
Writing the Analytics Memo and Corporate Brief
Group assignment. The memo and the brief as documents for two readers; leading with the finding in the brief; replacing adjectives with measurements; the so-what test; the group-work rules on documentation and contributions.One analysis, two documents, two readers
The group assignment asks a group to work with a business dataset, develop a clear research question, run a reproducible analysis using descriptive, causal or predictive methods, and communicate the findings twice: an Analytics Team Memo written for peers or a manager who will check the work, and a short Corporate Brief written for an executive who has to decide something.
The brief is not a shortened memo
The memo can begin with the question and work forwards to the finding.
- The memo and the brief as documents for two different readers
- Leading with the finding and the recommended action in the brief
- Replacing vague adjectives with the measurement they stand in for
- Applying the so-what test to every sentence in a draft
- The group-work rules on documentation, channels and contributions
It takes a technical, code-based approach, working through the fundamentals of structuring, transforming, visualising, analysing and interpreting data in R, and applies those skills across finance, accounting, marketing, economics and supply chain management.
No prior programming experience is expected.
Three dimensions of analytics
The subject organises the field into descriptive analytics, which establishes what happened; predictive analytics, which anticipates what happens next; and causal analytics, which uncovers cause-and-effect relationships.
Hands-on case studies apply these to real business problems, and the subject places the ethics of working with sensitive information inside the method rather than beside it: recognising and mitigating bias, balancing insight against individual rights, and communicating findings transparently.
How the subject is taught
Lectures run in weeks 1 to 12 and are built as live demonstrations rather than displays of finished code: a business question is translated into a plan, the plan is implemented, and results are checked for plausibility.
Workshops begin in week 2 and run as collaborative coding sessions in which groups work through prepared questions and report back.
Everything runs in Posit Cloud, and the prescribed text is R for Data Science (second edition).
What this guide covers
Thirteen chapters follow the twelve lecture weeks, from the three kinds of analytics question through reproducible workflows, visualisation, wrangling, reshaping and joining, to causal and predictive analytics and the two written products of the group assignment, with a drill chapter and a final-week plan.
Current deadlines, venues and any changes are published on the subject site.
Worked example · free
Reading a table of group averages without overreaching
- 1State what the table does support.
- 1Give the two separate reasons it cannot answer the question asked.
- 1Name the next analytical step.
Key terms
- Business analytics
- The practice of turning a business decision into a question a dataset can answer, and reporting the answer with its limits attached.
- Descriptive analytics
- Summarising what already happened in a record, using only rows that exist.
- Predictive analytics
- Anticipating a value that has not been observed yet, judged on data the rule never saw.
- Causal analytics
- Establishing whether one thing produced another, which requires a comparison with what would otherwise have happened.
- Grain
- What a single row of a table represents, such as one customer, one order or one firm-year.
- Tidy data
- A table in which each column is a variable, each row is an observation and each cell holds one value.
- Pipe operator
- A connector that passes the result on its left into the first argument of the function on its right.
- Grouped summary
- A table reduced to one row per group, each carrying statistics computed over that group.
- Join key
- The column or set of columns shared by two tables that identifies what a row represents in both.
- Panel data
- Observations on many units across many periods, one row per unit and period.
- Random assignment
- Allocating units to treatment or control by chance, so unmeasured differences spread evenly across both.
- Corporate brief
- The short document aimed at an executive who has to decide something on the strength of a finding.
CMCE10002 FAQ
How is the subject assessed, and is any component a pass condition?
Four components are published: in-class quizzes during workshops worth 10 per cent across weeks 4 to 12, a mid semester test worth 20 per cent in week 6, a group assignment worth 20 per cent due in week 12, and a two-hour end of semester examination worth 50 per cent. The subject guide states that successful completion requires a pass in the final exam, and that condition is printed with the examination row alone.
What does the mid semester test cover and how is it sat?
It runs for one hour with no reading time and covers material from the first five weeks. It is sat in person as a Canvas quiz using the Respondus LockDown browser, and mixes multiple choice with short answer, including questions that ask students to write and interpret code. Installing and testing the browser beforehand is the students' own responsibility.
What is the structure of the final examination?
Two hours, hand written, drawing on lectures, workshops and the assigned readings. The subject guide describes 20 multiple-choice questions carrying 40 per cent of the paper, alongside roughly four short-answer questions that carry subparts and the remaining 60 per cent, some of which may ask students to write code on paper and say what it does.
What does the group assignment ask students to produce?
An analytics project on a business dataset with a clear research question and a reproducible analysis using descriptive, causal or predictive methods. The findings are communicated in two formats: an Analytics Team Memo, written for a peer or manager who will check the work, and a short Corporate Brief, written for an executive who has to decide something.
Which software is used, and does anything need to be installed?
All exercises and assignments run in R and RStudio through Posit Cloud, a browser-based environment that requires no installation and saves work online so it can be reopened from any device. The University covers the usage fee associated with the subject.
Is prior programming experience needed to take this subject?
It is not expected. The subject describes itself as technical while stating that students with basic numeracy, curiosity about how data and technology are changing business, and a willingness to engage in structured problem-solving are well positioned to succeed.
How may generative tools be used across the semester?
The subject runs a learn-first policy. Before the mid semester test these tools may be used to explain concepts but not to generate or debug code. After it, code use is permitted with documentation, which means citing the tool, including the prompts and output, and explaining in your own words what the generated code does. They are not permitted at any point during the quizzes or the mid semester test.
Questions students ask in each CMCE10002 chapter
- What are the three kinds of analytics question?
- The subject names descriptive analytics for what happened in the past, predictive analytics for what happens next, and causal analytics for why something happened. Each demands different evidence, so naming the type first decides what a good answer must contain. Chapter 01 · What Business Analytics Is: Three Kinds of Question
- Do students need to install R for this subject?
- No local installation is required. All class exercises and assignments run through Posit Cloud in a web browser, the work is saved online so it can be reopened from any device, and the University covers the usage fee associated with the subject. Chapter 02 · Reproducible Workflows in Posit Cloud and Quarto
- What is the difference between a row count and a count of customers?
- A row count counts records. If the table holds one row per review or per order, a customer who appears several times is counted several times. Counting unique customer identifiers gives the number of people, and the gap between the two figures measures how much repetition the table carries. Chapter 03 · Reading R Code: Objects, Tibbles and Subsetting
- Why does naming the data alone produce an empty plot?
- Because the plot has been told which table to draw from and not yet what to draw. The geometry is what puts marks on the canvas, so an empty frame confirms the first step worked rather than signalling an error. Chapter 04 · Visualising Business Data with ggplot2
- Can two datasets have the same average and still look completely different?
- Yes, and the week 3 lecture opens with exactly that demonstration. Several datasets can be constructed to share a mean and a standard deviation on both axes while their scatterplots have nothing in common, which is why the subject plots before it summarises. Chapter 05 · Why Summary Statistics Alone Can Mislead
- Which comes first, filtering or ranking?
- Filtering. Ranking a table that still contains rows with missing or impossible values puts those rows exactly where a reader looks, which is the top of a sorted table. Removing them first is cheaper than explaining them afterwards. Chapter 06 · Wrangling One Table with dplyr
- Does grouping change the table?
- Nothing visible changes. Row for row the table is what it was, in the order it was, holding the values it held, and the only difference is a label naming each row's group. What the label does is change how the next operation behaves, which is why its effect appears two lines later. Chapter 07 · The Pipe and Grouped Business Summaries
- What makes a dataset tidy?
- Three conditions together: each column is a variable, each row is an observation and each cell holds one value. The definition is a technical contract rather than a description of neatness, and the tools in this subject assume it has been honoured. Chapter 08 · Tidy Data and Reshaping with Pivots
- Which join keeps every row of the first table?
- The left join retains every row on its first side and brings columns across from the second wherever a key matched. Rows that found no partner survive with empty cells in the attached columns, which is the join working correctly rather than an error. Chapter 09 · Joining Tables on a Shared Key
- What is the difference between a cross section and a panel?
- A cross section holds many units at one moment, so one row is one unit. A panel holds many units across many periods, so one row is one unit in one period. The panel can answer questions about change within a unit that a cross section cannot reach at all. Chapter 10 · Variation in Cross Section, Time Series and Panel Data
- Why is a before-and-after comparison weak evidence?
- Because everything else that changed between the two periods is absorbed into the difference: seasonality, campaigns, news events and competitor behaviour among them. The observed change is the true effect plus all of those, and nothing in the data separates them. Chapter 11 · Causal Analytics and Randomised Experiments
- Why can a forecast not be scored on the data it was built from?
- Because any sufficiently flexible rule can be made to trace rows it has already seen. Scoring on those rows measures how closely it reproduces the past, not whether it will be right about a period nobody has observed yet. Chapter 12 · Predictive Analytics, Stored Data and APIs
- What two documents does the group assignment require?
- An Analytics Team Memo, written for a peer or manager who will check the work, and a short Corporate Brief, written for an executive who has to decide something. The subject frames them as mirroring the way findings actually move through an organisation. Chapter 13 · Writing the Analytics Memo and Corporate Brief
How to study for the exam
Both timed assessments in this subject are sat without a console, so the examinable skill is predicting what a table looks like after an operation rather than running one and reading the result. Build revision around that: cover the output, write down the row and column count you expect, then run the code and compare.
Work the weekly question sets, because they were designed to prepare both the multiple-choice and the short-answer components. In the final fortnight, sit the first mock paper cold and timed, sort its errors into recall failures and reasoning failures, and give the last three teaching weeks one deliberate pass, since nothing before the examination tests them.
Your AI Statistics tutor for CMCE10002
Stuck on a hard CMCE10002 question? Sia is AskSia’s AI Statistics tutor — ask any CMCE10002 Foundations of Business Analytics question and get a clear, step-by-step explanation grounded in how the course is actually taught and assessed. Read this whole study guide free, then take your hardest questions to Sia.
