AU-03 · full course

Data Analysis and Data Science Practice

Ask a clear question, find and clean data, describe and chart it, test claims and report honestly. You practise on public or practice data and learn to state limits openly.

about this course

What you will be able to do

L4 Practitioner8 modules · 150 h guidedPractice 70% · Share 20% · Study 10%

Ask a clear question, find and clean data, describe and chart it, test claims and report honestly. You practise on public or practice data and learn to state limits openly.

Understand Explain how a question, data and a decision connect.
Apply Clean and describe a dataset with a full change log.
Analyse Test whether a pattern is real or due to chance.
Create Create a clear chart and an honest report.

Task, condition and standard

Task

Answer one real question using a public or practice dataset. Produce a cleaned dataset with a change log, a summary of the numbers, two honest charts, a test of one claim and a one-page report that states limits.

Condition

About 150 hours over 3 to 6 months. You need a computer, a free spreadsheet program and notebook tools, practice data, a mentor and a peer circle.

Standard

Clear measurable question; data dictionary; cleaned data with change log and raw copy kept; summary statistics; two charts with labelled units; one tested claim; one-page report with limits; at least 70% on the final check.

Learning only. Use public or practice data; never publish personal data. Check with your local authority about privacy rules before using data about people.

Requirements: none. Who it is for: learners 16 and over, workers changing field, small-business starters, teams. Currency is shown on every number.

lesson by lesson

8 modules with worked examples (150 h guided)

Module 1 Ask a good question 12 hUnderstand

A clear question saves hours of wandering through data. It names who will use the answer, what they will decide and how the result is measured.

Start by naming the person who will use the answer and the decision they will make.
Make the question measurable by naming the exact number, such as average daily sales in dollars.
State the time period and place, such as the last 12 months in one region.
Write three possible questions and rank them by how useful and how answerable they are.
Check that the data you need exists before you commit to the question.

Worked example. A cafe owner asks: do weekday sales fall in the 3 days after a promotion ends? That is measurable, has a time frame and links to a decision about when to run the next promotion. The vague version, how are sales going, could not be answered.

Activity. Write three candidate questions, rank them and choose one, giving a reason for the choice.

Standard to meet. Choose a question that names a user, a decision, a measure and a time period.

Module 2 Get and understand data 16 hApply

Before analysing anything you must know what each column means, where the data came from and what is missing. Skipping this step is the most common cause of wrong answers.

Record the source, the date it was collected and the date you downloaded it.
Write a data dictionary listing each column, its meaning, its unit and its type, such as number, date or text.
Check units carefully, because a column mixing dollars and cents or kilograms and grams gives wrong totals.
Count missing values in every column and find how they are coded, for example blank, zero or 999.
Look at the first 20 rows and the last 20 rows to spot odd entries.

Worked example. A sales table has 400 rows. The age column has 37 entries of 999, which is a code for unknown, not a real age. If you included them the average age of 400 customers would be inflated, so they must be treated as missing.

Activity. Describe your dataset with a data dictionary and a count of missing values for each column.

Standard to meet. Give every column a meaning, a unit and a missing count, and identify any coded values.

Module 3 Clean data 24 hApply

Cleaning makes the data fit for analysis without secretly changing the story. The rule is to keep the raw copy untouched and record every change.

Save a read-only raw copy first and work only on a second copy.
Find and remove exact duplicate rows, and record how many you removed.
Fix obvious entry errors, such as 2024 typed as 2042, only when you can show why it is wrong.
Decide how to treat missing values, either remove the rows, fill with a stated value or leave them blank, and write down the reason.
Keep a change log with the date, the action, the number of rows affected and the reason.

Worked example. A file starts with 412 rows. After removing 12 duplicates it has 400 rows. The change log reads: removed 12 exact duplicate rows, 412 to 400, because the same order was entered twice on import. The raw file remains saved unchanged.

Activity. Clean your dataset and submit the cleaned file, the raw file and a change log with at least five entries.

Standard to meet. Check that row counts before and after each step match, and keep the raw file unchanged.

Module 4 Describe 20 hAnalyse

Summary numbers show the centre and spread of the data. Choosing the right summary stops a few extreme values from misleading you.

The mean is the sum divided by the count, and it is pulled by very large or very small values.
The median is the middle value when sorted, and it is steadier when data is skewed.
Spread can be shown with the minimum, maximum and range, or with the middle half of values.
When comparing groups, report counts alongside percentages so readers can see how big each group is.
Group the data by one category at a time and compare the same measure.

Worked example. Five households spend A$20, A$22, A$25, A$30 and A$163 on a weekly item. The mean is (20+22+25+30+163) / 5 = A$52, but the median is A$25, which describes the typical household better because one very high value pulls the mean up.

Activity. Compute the mean, median, minimum and maximum for two columns and compare two groups with counts and percentages.

Standard to meet. Calculate all numbers correctly, and explain in one sentence which summary suits each column.

Module 5 Visualise 20 hCreate

A chart should answer your question at a glance. The chart type, the labels and the axis starting point decide whether it informs or misleads.

Use a bar chart to compare categories and a line chart to show change over time.
Label both axes with the measure and the unit, and give the chart a title that states the finding.
Start a bar chart axis at zero, because a cut axis makes small differences look huge.
Use few colours and use the same colour for the same thing across charts.
Test the chart on someone new and check they can state the main point in about 10 seconds.

Worked example. A bar chart compares weekly sales of A$4,200 and A$4,500 for two stores. With the axis starting at zero the bars look similar. If the axis starts at A$4,000 one bar looks two and a half times taller, which is misleading, so zero is used and the title reads: store B sold 7% more.

Activity. Create two charts, one for categories and one for time, each with a clear title, labelled axes and units.

Standard to meet. Check that a reader can state the main point of each chart in 10 seconds, and start the axes honestly.

Module 6 Test claims 24 hAnalyse

A pattern in a sample might be real or might be chance. Testing means checking size, fairness of comparison and alternative explanations before you claim anything.

Check the sample size, because a pattern in 8 cases is much weaker than the same pattern in 800.
Compare like with like, for example the same weeks of the year or the same type of customer.
Calculate the difference between groups and note how large it is relative to normal variation.
Remember that two things rising together does not show one causes the other.
Look for a third factor that could explain both, such as hot weather.

Worked example. Ice cream sales and sunburn cases both rise in summer, but ice cream does not cause sunburn; hot sunny weather drives both. If group A averages 52 and group B averages 48 in samples of only 6 each, a 4-point gap could easily be chance.

Activity. Choose one claim from your dataset and test it, writing a short note on sample size, fair comparison and other explanations.

Standard to meet. State the sample sizes, the size of the difference and at least one alternative explanation in the note.

Module 7 Ethics and privacy 16 hEvaluate

Data about people can harm them if handled carelessly. Ethical analysis protects individuals and checks that results are fair to groups.

Remove names, addresses and other identifiers before sharing or publishing any data.
Check small groups, because a group of 3 people in a table can make someone identifiable.
Compare results across groups such as age bands or regions to see whether one group is treated differently.
Use data only for the purpose it was collected for unless people agreed to more.
Check with your local authority for the privacy rules in your area before using data about people.

Worked example. A table shows average spend by suburb, and one suburb has only 2 customers. Publishing it could reveal their spending, so the 2 are merged into a larger region of 31 customers. A bias check shows approval rates are 70% for one age group and 52% for another, which is flagged for review.

Activity. Write a one-page note covering identifiers removed, small groups handled and one bias check on your data.

Standard to meet. List the steps taken in the note, show one group comparison with numbers and name any remaining risk.

Module 8 Report 18 hCreate

A report is useful only if the reader can find the answer quickly and trust how you got it. Lead with the answer, then show the evidence and the limits.

Put the answer in the first sentence, written in plain words with the key number.
Follow with the evidence, using one or two charts and a short table.
State limits openly, such as a small sample, missing values or a short time period.
Finish with a recommended next step and who should take it.
List the steps in order so another person could repeat your work from the raw file.

Worked example. Report opening: weekday sales fall 11% in the 3 days after a promotion, from A$3,200 to A$2,850 a day. Evidence is one chart and one table. Limits: only 6 promotions were recorded and no holiday weeks were checked. Next step: test a promotion with a follow-up offer.

Activity. Write and present a one-page report with the answer, evidence, limits, next step and a short list of steps to repeat the work.

Standard to meet. Put the answer first in the report, show evidence with units and state at least two limits and one next step.

assessment

How you are checked

Evidence: your records, numbers and reflections. Two peers and one mentor mark it with this rubric. Pass: no criterion at "Not yet" and at least 70% on the check.

Marking rubric
Criterion	Not yet	Meets	Exceeds
Evidence quality	Claims without records	Records for most claims	Every claim linked to a dated record
Numbers and units	Missing units or currency	Correct, units shown	Correct, with a check calculation
Rules and safety	Rules ignored	Rules named and followed	Rules checked with the local authority, note kept
User focus	No real user	One real user	Several users, changes made from feedback
Reflection	No lessons written	Three lessons written	Lessons turned into a changed plan with dates

Ten-question check, with model answers

Q1. Name one trait of a good analysis question.

Model answer: It is measurable and tied to a decision someone needs to make. (Covered in module 1: Ask a good question. To pass that module: Choose a question that names a user, a decision, a measure and a time period.)

Q2. Why must you check units before analysing?

Model answer: Mixed units, such as dollars and cents, give wrong totals and wrong averages. (Covered in module 2: Get and understand data. To pass that module: Give every column a meaning, a unit and a missing count, and identify any coded values.)

Q3. Why keep an untouched raw copy of the data?

Model answer: So you can check or undo any change and show how you reached your results. (Covered in module 3: Clean data. To pass that module: Check that row counts before and after each step match, and keep the raw file unchanged.)

Q4. When is the median a better summary than the mean?

Model answer: When the data is skewed or has extreme values, because the median is not pulled by them. (Covered in module 4: Describe. To pass that module: Calculate all numbers correctly, and explain in one sentence which summary suits each column.)

Q5. Why must every chart axis have a label and a unit?

Model answer: Readers cannot interpret the numbers without knowing what is measured and in which unit. (Covered in module 5: Visualise. To pass that module: Check that a reader can state the main point of each chart in 10 seconds, and start the axes honestly.)

Q6. Does a correlation between two things prove one causes the other?

Model answer: No, a third factor or chance could explain the link, so you need further evidence. (Covered in module 6: Test claims. To pass that module: State the sample sizes, the size of the difference and at least one alternative explanation in the note.)

Q7. Name one check for bias in data.

Model answer: Compare the results across groups, such as age bands, and look for large unexplained gaps. (Covered in module 7: Ethics and privacy. To pass that module: List the steps taken in the note, show one group comparison with numbers and name any remaining risk.)

Q8. In what order should a report be laid out?

Model answer: Answer first, then evidence, limits and next step. (Covered in module 8: Report. To pass that module: Put the answer first in the report, show evidence with units and state at least two limits and one next step.)

Q9. What is the task of this course, in one sentence?

Model answer: Answer one real question using a public or practice dataset. Produce a cleaned dataset with a change log, a summary of the numbers, two honest charts, a test of one claim and a one-page report that states limits. (Covered in: Task, condition and standard, near the top of this page.)

Q10. What standard must your final evidence meet?

Model answer: Clear measurable question; data dictionary; cleaned data with change log and raw copy kept; summary statistics; two charts with labelled units; one tested claim; one-page report with limits; at least 70% on the final check. (Covered in: Task, condition and standard, near the top of this page.)

where next

Ladder L1 to L7, slide by slide with notes

Slide 1: L1 Basic, First words, tools and safety. Speaker note: Learner can say what Data Analysis and Data Science Practice is about and name the safety rule. Use module 1, Ask a good question, as the first short unit (about 2 to 6 hours). Say this rule aloud: Learning only. Use public or practice data; never publish personal data. Check with your local authority about privacy rules before using data about people. No record is needed at this level.
Slide 2: L2 Awareness, What the field is and who it serves. Speaker note: Learner explains the idea in plain words: Explain how a question, data and a decision connect. Shared with schools and community groups. Note for the guide: ask one learner to explain it to a neighbour and check the words are right.
Slide 3: L3 Foundation, Guided tasks with a checklist. Speaker note: Learner does guided tasks from the first three modules (Ask a good question; Get and understand data; Clean data) with a checklist. Evidence is a small set of dated records. Note for the guide: sit beside the learner for the first record, then step back.
Slide 4: L4 Practitioner, Run the full task alone (this course). Speaker note: Task: Answer one real question using a public or practice dataset. Produce a cleaned dataset with a change log, a summary of the numbers, two honest charts, a test of one claim and a one-page report that states limits. Standard: Clear measurable question; data dictionary; cleaned data with change log and raw copy kept; summary statistics; two charts with labelled units; one tested claim; one-page report with limits; at least 70% on the final check. Note for the guide: a mentor reviews the records; two peers mark with the rubric above.
Slide 5: L5 Specialist, One narrow area to depth. Speaker note: Learner picks one area, for example Clean data (the longest module at 24 h), and goes deeper than this course asks. Learner then teaches a peer circle of three or four. Note for the guide: check the teaching, not only the work.
Slide 6: L6 Advanced, Lead a team and review evidence. Speaker note: Learner leads a small team or designs a service in this field and marks other people's evidence against the rubric. Roles this points to: Data analyst assistant; Reporting analyst; Junior data scientist. Note for the guide: sample three records from the team and compare marks.
Slide 7: L7 Super expert, Set and audit the standard. Speaker note: Learner writes the method for Data Analysis and Data Science Practice as a short guide, audits other learners' evidence, and publishes worked examples back to learners. Note for the guide: worked examples keep currency and units on every number and are checked by a second expert.

Notes: credits stack and you can step in and out. This course is L4. Aligned with, not accredited by, national credit frameworks.

jobs this leads to

Where this takes you

Data analyst assistant
Reporting analyst
Junior data scientist
Research assistant

ten locations and global

Your place

Local custom changes rules, language, currency and who you practise with. Check each rule with your local authority before you act.

India (Bengaluru)

Language: English, Hindi, Kannada. Currency: INR. Data rule for what you publish: Digital Personal Data Protection Act 2023. Skills frame to cite: National Skills Qualifications Framework (NSQF) and the national credit framework.

Netherlands (Eindhoven area)

Language: Dutch, English. Currency: EUR. Data rule for what you publish: EU GDPR (Dutch: AVG), supervised by the Autoriteit Persoonsgegevens. Skills frame to cite: the Dutch qualification framework (NLQF) linked to the European Qualifications Framework.

UAE (Dubai and Abu Dhabi)

Language: Arabic, English. Currency: AED. Data rule for what you publish: Federal Decree-Law No. 45 of 2021 (personal data protection). Skills frame to cite: the emirate education and training regulators (for example KHDA in Dubai) and the national qualifications framework.

Brazil (Sao Paulo)

Language: Portuguese (Spanish and English support). Currency: BRL. Data rule for what you publish: LGPD (Lei Geral de Protecao de Dados). Skills frame to cite: the national technical-education catalogue and the national S-system training bodies.

Canada (Toronto)

Language: English, French. Currency: CAD. Data rule for what you publish: PIPEDA, with provincial rules where they apply. Skills frame to cite: the provincial colleges framework and the national credential-recognition guidance.

United States

Language: English, Spanish. Currency: USD. Data rule for what you publish: state privacy laws (for example California), and COPPA for children under 13. Skills frame to cite: state and community-college credit systems.

Singapore

Language: English, Mandarin, Malay, Tamil. Currency: SGD. Data rule for what you publish: Personal Data Protection Act (PDPA). Skills frame to cite: the Singapore national skills frameworks and the national credit system.

Australia

Language: English. Currency: AUD. Data rule for what you publish: the Privacy Act 1988 and the Australian Privacy Principles. Skills frame to cite: the Australian Qualifications Framework (AQF) levels.

Hong Kong (Eastern Asia)

Language: Chinese (Cantonese, Mandarin), English. Currency: HKD. Data rule for what you publish: the Personal Data (Privacy) Ordinance. Skills frame to cite: the Hong Kong Qualifications Framework.

Georgia and Israel (Western Asia)

Language: Georgian, Hebrew, Arabic, English. Currency: GEL or ILS. Data rule for what you publish: the personal data law of the country where the learner lives. Skills frame to cite: the national qualifications framework of that country.

Cayman Islands (Caribbean)

Language: English. Currency: KYD. Data rule for what you publish: the Data Protection Act. Skills frame to cite: the local education and training framework.

Mexico

Language: Spanish, English. Currency: MXN. Data rule for what you publish: the federal law on personal data held by private parties. Skills frame to cite: the national competency certification system and the education ministry levels.

Draft

Draft flagship-depth course. Own wording from the planning booklet; no third-party text. Not live.