FIT1043 Introduction to Data Science
The Complete Exam Bible · S2 2026FIT1043 Overview
- Monash University
- S2 2026
- 11 chapters
- Information Technology
FIT1043 Introduction to Data Science is Monash University's first-year, introductory Faculty of Information Technology unit that walks through the whole data-science lifecycle — from understanding what data science is and where it sits (Drew Conway's diagram and the 'danger zone'), through acquiring, wrangling and visualising data, into predictive modelling (regression, then classification and clustering), and on to Big Data storage/processing and data governance.
- Assessed by Test 1 (Quiz / Test) 10% · Data Science Assignment 1 20% · Data Science Assignment 2 20% · Final Examination 50%
- Key terms Data science lifecycle, Drew Conway 'danger zone', Supervised vs unsupervised learning, The four V's
- How to prepare Treat FIT1043 as a definitions-and-methods unit and build a one-page recall sheet per week rather than re-reading slides.
- Most asked Where can I find past exam papers / practice for FIT1043?
How FIT1043 is assessed
| Component | Weight | Format |
|---|---|---|
| Test 1 (Quiz / Test) | 10% | On-campus eAssessment with online supervision, ~Week 5; 5 MCQ + 10 short-answer, closed book, ~1h10m (incl. 10 min reading); examines Weeks 1-4 |
| Data Science Assignment 1 | 20% | Individual predictive-analytics exercise in Python/Jupyter (read/describe data, train/test, model evaluation, k-means), due ~Week 10 |
| Data Science Assignment 2 | 20% | Individual BASH-shell processing of a large file -> CSV -> analysis/visualisation in R, due ~Week 12 |
| Final Examination | 50% | Supervised closed-book eExam, 2h10m, 65 marks; Part 1 = 15 MCQ (1 mark each), Part 2 = 25 short-answer (2 marks each); whole unit |
What FIT1043 covers
FIT1043 Introduction to Data Science is assessed through three in-semester threshold tasks — Test 1 (10%), Data Science Assignment 1 in Python/Jupyter (20%) and Assignment 2 in BASH + R (20%) — capped by a closed-book final eExam worth 50% of the unit.
This eleven-chapter map follows the Monash teaching schedule from what data science is, through Python and R, wrangling, visualisation, regression, classification and clustering, into Big Data and governance. Use it to see how each week's concept turns into a one-mark MCQ or a two-mark short-answer on the final.
Data Science and Data in Society
What data science is · the danger zone · lifecycle · roles · data value chain (Week 1)Week 1 of Monash FIT1043 Introduction to Data Science frames what data science actually is — the work of turning data into insight beyond the core task of data analysis — using Drew Conway's diagram of hacking skills, maths/statistics and domain expertise, and its cautionary 'danger zone'.
- What data science is: working with data beyond core data analysis, and how it differs from data engineering and data analysis (the boundary is NOT fixed)
- Drew Conway's diagram: hacking/programming skills, maths & statistics knowledge, substantive/domain expertise — and data science at their intersection
- The 'danger zone': hacking skill + domain expertise WITHOUT maths/statistics — running end-to-end ML and reporting coefficients you cannot interpret
- When machine learning is useful: expertise unavailable, humans cannot state the rules, or humans are expensive to use
- The data-science lifecycle and the roles in a project: statistician, analyst, archivist, systems architect
- The Standard Value Chain — Collect, Wrangle, Analyse, Present — as the middle pipeline of the lifecycle
Introduction to Python for Data Science
Python basics · pandas DataFrames · groupby/select · Jupyter (Week 2)Week 2 of Monash FIT1043 Introduction to Data Science puts Python and Jupyter notebooks to work as the unit's main analysis tool, centred on the pandas library: building and inspecting DataFrames, selecting and filtering rows, and grouping and aggregating.
- Python as the most popular data-analytics language; coding in Jupyter notebooks (.ipynb = Interactive Python NoteBook)
- pandas DataFrame: a structure that stores tabular data; head(), tail(), sample(), .shape, .dtypes, .describe()
- Indexing and selection: df['col'], df.loc[...], boolean masks
- Boolean filtering and operator precedence: combining conditions with &, | and parentheses (the parenthesis trap)
- Derived columns: df['Total'] = df['Math'] + df['English']; averages from derived columns
- Reshaping and merging: pd.melt (wide to long); pd.merge on a key — inner join drops unmatched rows vs how='left' keeping all left rows
Data Sources & Data Wrangling
Data types/formats · sources & APIs · cleaning · tidy data (Week 3)Week 3 of Monash FIT1043 Introduction to Data Science covers where data comes from — structured vs unstructured data, open data and linked open data, APIs and scraping — and how to make dirty data usable through wrangling and cleaning.
- Kinds and sources of data: structured vs unstructured; open data and Linked Open Data (machine-readable, public)
- APIs as data sources: Google Maps, Twitter, Facebook, OpenWeather; how different APIs are accessed
- Data-quality problems: impossible values, invalid codes, inconsistent formats; inspect issues and recommend fixes
- Handling missing data: replace with a special 'unknown' value, replace with an average, or drop the row/column — all can be valid
- Wrangling in pandas: fixed-width read (read_fwf), assigning column names, standardising values with .loc, dropping bad rows with boolean filters
- Tidy-data principles: each variable a column, each observation a row
Data Visualisation & Descriptive Statistics
Chart types · mean/median/variance · distributions · choosing a visualisation (Week 4)Week 4 of Monash FIT1043 Introduction to Data Science pairs descriptive statistics — mean, median, mode, range, standard deviation and quartiles — with choosing an appropriate visualisation (line, histogram, boxplot, bar, pie, scatter).
- Why visualise: get a feel for the data before and alongside numeric summary
- Descriptive statistics: mean, median, mode; range, variance, standard deviation; quartiles (25%/50%/75%)
- Choosing a chart: line plot (trend), histogram (distribution of one variable), boxplot (five-number summary + outliers)
- Bar chart (compare categories), pie chart (parts of a whole), scatter plot (relationship between two variables)
- Reading a boxplot: median line, interquartile-range box, whiskers and outliers; grouping by a category
- Histograms as a special kind of bar chart; bins control granularity
Data Analysis Theory
Predictive vs descriptive · supervised/unsupervised · linear regression intro · train/test (Week 5)Week 5 of Monash FIT1043 Introduction to Data Science steps back to the theory of analysis: the difference between predictive and descriptive modelling, supervised vs unsupervised learning, and the idea of a model with features (independent variables) predicting a target (dependent variable).
- Machine learning: analysing data to perform a task from patterns and inference rather than explicit instructions; pattern detection is central
- Predictive vs descriptive modelling; the general pattern vs outlying points
- Supervised vs unsupervised learning; features (independent/predictor) vs target (dependent/response) variable
- Linear regression: fitting y = slope * x + intercept to detect a linear relationship
- Reading the fit: slope (direction/rate), intercept, and r (correlation) — magnitude near 1 or -1 = strong, sign = direction
- Python: linregress(x, y) returning slope, intercept, r_value, p_value, std_err; predicted line = [slope*xi + intercept]
Regression Analysis
Fitting/interpreting regression · overfitting & underfitting · bias-variance · ensembles (Week 6)Week 6 of Monash FIT1043 Introduction to Data Science deepens regression into the core modelling trade-off of the unit: bias vs variance, and its symptoms overfitting and underfitting.
- Parameter estimation: fitting model coefficients, e.g. by least squares (best fit = least squared error over the x-range)
- Model family and hyper-parameters: a family is a class of models set by hyper-parameters (here the polynomial ORDER); a member is fixed by its coefficients
- Bias: how closely a member of the family CAN fit the truth — simple/low-order = large bias, complex/high-order = small bias
- Variance: the spread between fitted curves across samples vs the best possible (infinite-data) fit
- The bias-variance trade-off: underfit (too simple) vs overfit (too complex, erratic on finite data)
- Overfitting vs underfitting: a high-order poly with about as many coefficients as data points 'goes wild'
Classification & Clustering
Classification vs regression · decision trees · k-means · confusion matrix (Week 7)Week 7 of Monash FIT1043 Introduction to Data Science covers the two most examinable modelling families: classification (a categorical target) versus regression (a continuous target), and clustering (unsupervised grouping).
- Classification vs regression: categorical/discrete target (spam vs not-spam) vs continuous target (price, salary)
- Decision trees and regression trees: how splits partition the data (concept)
- Random forest: an ensemble of decision trees (concept)
- k-means clustering: unsupervised grouping of similar points into k clusters
- The k-means algorithm: define K, initialise centroids, assign points to nearest centroid, update centroids, repeat until centroids stop changing significantly
- Confusion matrix: TP, FP, FN, TN and accuracy = (TP+TN)/total
Introduction to R for Data Science
R basics · vectors/data frames · dplyr · plotting in R (Week 8)Week 8 of Monash FIT1043 Introduction to Data Science switches tools to R and RStudio, the unit's second analysis language, and repeats the earlier skills in R idiom: vectors and data frames, summary statistics, reading large/compressed files, fitting a linear model with lm(), and plotting.
- R and RStudio / RStudio Cloud; R as the second most popular data-analytics language
- R syntax basics: assignment with <-, vectorised operations (element-wise), c() to build vectors, sequences 1:10
- Random data generation: runif (uniform), rnorm (normal), sample()
- Summary statistics: mean(), max(), min(), sd(), summary(), quantile(), median and quartiles
- Data frames in R: data.frame(...), rows = observations, columns = variables; head(), tail(), str(), unique(); column selection with $ and [ , ]
- Reading large/compressed files: read.table(gzfile('...csv.gz'), header=TRUE, sep=',')
Characterising Data & 'Big' Data
The 4 V's · growth laws · BASH/shell for large files (Week 9)Week 9 of Monash FIT1043 Introduction to Data Science defines what makes data 'big' through the V's — Volume, Velocity, Variety, Veracity (and often Value) — and connects them to the growth laws (Moore's, Koomey's, Bell's, Zimmerman's) and the types of metadata.
- Big data defined: any attribute (among the V's) that challenges the constraints of a system's capability or a business need
- Volume (size of data) and Velocity (frequency/pace of incoming data to process)
- Variety (different types of data) and Veracity (accuracy/truthfulness/reliability; uncertainty of data)
- Value as a common fifth V; the original 3 V's as the basis for further V's
- Growth laws: Moore's Law, Koomey's Law, Bell's Law, Zimmerman's Law
- Metadata types: descriptive (title/author), structural (chapters/XML elements/MPEG containers), administrative (version number/archiving date/DRM)
Big Data Processing
SQL vs NoSQL · distributed processing · MapReduce · Hadoop vs Spark (Week 10)Week 10 of Monash FIT1043 Introduction to Data Science is about storing and processing data too big for one machine: SQL vs NoSQL databases, distributed processing, the MapReduce paradigm, and the contrast between Hadoop (disk-based HDFS + MapReduce) and Spark (in-memory).
- Database types: characterising different kinds of database
- SQL vs NoSQL: structured relational tables vs flexible non-relational stores, and when each suits the data
- Distributed processing: spreading computation and storage across a cluster of machines
- The MapReduce paradigm: a map step that emits key-value pairs and a reduce step that aggregates them
- The canonical MapReduce word-count example
- Hadoop: a system for partitioning computation across a compute cluster (HDFS storage + MapReduce)
Data Governance, Ethics & Privacy
Data lifecycle vs value chain · privacy & sensitive data · ethics & security (Week 11)Week 11 of Monash FIT1043 Introduction to Data Science closes the unit with governance: the internal data lifecycle versus the external data value chain, the tension between business and legal objectives, and the relationship between ethics, privacy, storage, security and analysis.
- Data management from two perspectives: internal = the data lifecycle; external = the data value chain
- Differentiating the requirements of the lifecycle (internal) from the value chain (external)
- Conflicting business vs legal objectives in managing data
- The relationship between ethics, privacy, storage, security and analysis
- Sensitive/personal information: what it is and why it must be kept confidential (e.g. emails contain addresses, phone numbers, financial info)
- Technological drivers of privacy loss: technology makes surveillance easier unless measures are put in place
It is a concepts-and-tools unit: you work in Python/Jupyter (Weeks 2-7), R/RStudio (Week 8) and the BASH shell (Weeks 9-12), and the marks reward crisp definitions, reading code correctly, and short justified answers rather than long essays.
Monash runs the assessment as three in-semester tasks that each carry a threshold hurdle — Test 1 (10%, an on-campus supervised eAssessment on Weeks 1-4), Data Science Assignment 1 (20%, a Python/Jupyter predictive-analytics exercise) and Data Science Assignment 2 (20%, BASH-shell processing of a large file then analysis in R) — capped by a closed-book final eExam worth 50% of the unit.
The final is a supervised eExam of 2 hours 10 minutes (65 marks): Part 1 is 15 multiple-choice questions (1 mark each) and Part 2 is 25 short-answer questions (2 marks each). Threshold hurdles apply: you must score at least 45% on the final scheduled assessment, at least 45% in total across the in-semester tasks, and reach an overall unit mark of at least 50% to pass — miss a hurdle and the unit is capped at NH (fail, max 45).
That structure feeds the Weighted Average Mark (WAM) later IT units build on, so steady weekly work beats a SWOTVAC cram. Confirm the current offering, exact dates and rules on the Monash Handbook and your unit's Moodle.
Worked example · free
Evaluating a classifier from a confusion matrix (accuracy, recall, precision)
- +1Accuracy = (TP + TN) / total = (30 + 140) / 200 = 170/200 = 0.85 = 85%. (Reads off the whole matrix: the fraction of all predictions that were correct.)
- +1Sensitivity (recall, true-positive rate) = TP / (TP + FN) = 30 / (30 + 10) = 30/40 = 0.75 = 75%. When the transaction really is fraud, the detector catches it 75% of the time.
- +1Specificity (true-negative rate) = TN / (TN + FP) = 140 / (140 + 20) = 140/160 = 0.875 = 87.5%. Precision = TP / (TP + FP) = 30 / (30 + 20) = 30/50 = 0.60 = 60%.
- +1A fraud detector most wants high sensitivity/recall: missing real fraud (a false negative) is the costly error, so you optimise TP/(TP+FN). (A spam filter, by contrast, fears false positives and leans on precision/specificity.)
Key terms
- Data science lifecycle
- The end-to-end process of a data-science project, from framing the problem and acquiring data through wrangling, analysis and communicating results. FIT1043's ULO 1 asks you to detail its phases and the roles involved.
- Drew Conway 'danger zone'
- The region of Conway's diagram where hacking/programming skill meets domain expertise WITHOUT maths/statistics — people who can run end-to-end machine learning and report coefficients but cannot interpret them, so their conclusions are unreliable.
- Supervised vs unsupervised learning
- Supervised learning trains on labelled examples to predict a target (classification for a categorical target, regression for a continuous one); unsupervised learning, such as k-means, finds structure with no target label.
- The four V's
- The attributes that make data 'big': Volume (size), Velocity (pace of incoming data), Variety (different types) and Veracity (accuracy/reliability). Big data = any V that challenges a system's capability or a business need; Value is a common fifth.
- Confusion matrix
- A 2x2 table of a classifier's true/false positives and negatives, from which accuracy = (TP+TN)/total, sensitivity/recall = TP/(TP+FN), specificity = TN/(TN+FP) and precision = TP/(TP+FP) are read.
- Bias-variance trade-off
- Simple/low-order models have large bias (cannot fit a complex truth) but small variance; complex/high-order models have small bias but can 'go wild', giving large variance. The best model balances the two — underfit vs overfit.
FIT1043 FAQ
Is FIT1043 hard?
It is broad rather than deep. FIT1043 is a first-year, introductory Faculty of Information Technology unit, so the maths stays at descriptive-statistics and read-the-slope-and-r level, but you cover a lot of ground — the data-science lifecycle, Python/pandas, wrangling, visualisation, regression, classification and clustering, R, Big Data and governance — and you switch between three toolsets (Jupyter, R, BASH).
Most students find the challenge is keeping every concept and definition straight and reading code correctly under time pressure, not the algebra. Doing the weekly quizzes and applied sessions as you go, rather than cramming through SWOTVAC, is what makes it manageable and protects your WAM.
Can AI help me with FIT1043?
Yes, as a study aid. Sia is an AI tutor trained on how FIT1043 is actually taught and assessed; it can explain the four V's, walk through the k-means steps, show why salary is a regression target and spam a classification target, or read a pandas groupby line with you, step by step, and check your reasoning on a practice question.
It explains the method and helps you understand it; it does NOT do graded assessment for you, and Monash University academic-integrity rules apply — so use it to learn and rehearse, not to produce work you submit.
Where can I find past exam papers / practice for FIT1043?
Start on your unit's Moodle: FIT1043 posts a sample exam, a sample solution and a Mock Exam as the official exam-prep artifacts, and the weekly quizzes (Weeks 2-10) and applied-session solutions are formative practice for both Test 1 and the final.
This guide also includes a re-authored practice exam that mirrors the eExam's shape — 15 one-mark MCQs plus short-answer definitions and code-reading — with fresh items, and you can ask Sia to generate more questions in the same style and explain each step. Treat any third-party 'model answers' with caution and confirm what is officially provided on Moodle and the Monash Handbook.
What are the hurdles and assessment rules in FIT1043?
FIT1043 has threshold hurdles.
To pass you must score at least 45% on the final scheduled assessment (the exam), at least 45% in total across the in-semester tasks, and reach an overall unit mark of at least 50% — miss any one and you receive a fail grade (NH) with the unit mark capped at 45. The in-semester tasks are Test 1 (10%), Assignment 1 (20%) and Assignment 2 (20%), each carrying the threshold-hurdle badge, and the final eExam is 50%.
Grade bands follow the Monash scale (HD >=80, D, C, P; NH = fail). Confirm the exact weights, dates and rules on the unit guide and Moodle.
Is the FIT1043 final open- or closed-book, and what's on it?
The final is a supervised, closed-book eExam of 2 hours 10 minutes worth 65 marks: Part 1 is 15 multiple-choice questions (1 mark each) and Part 2 is 25 short-answer questions (2 marks each), covering the whole unit. No notes, texts or websites are permitted. Short answers reward clear, simple English and bullet points are acceptable — extensive prose is not required.
You will not be asked to write code, but you may be asked to read a snippet and state its output. Confirm the current format, date and venue on the Monash Handbook and your Moodle before the day.
Questions students ask in each FIT1043 chapter
- What is the difference between data science, data engineering and data analysis?
- Data analysis is the core task of drawing insight from a prepared dataset; data engineering builds and maintains the pipelines and storage that supply the data; data science is the broader lifecycle around both, from framing the problem and acquiring data to communicating results. In FIT1043 the key exam point is that the boundary between the three is NOT fixed — roles overlap, and a data scientist often touches all three. Chapter 01 · Data Science and Data in Society
- Do I have to write Python code in the FIT1043 exam?
- No. The unit is explicit that you will not be asked to write code in Test 1 or the final. However, you may be asked to READ a snippet and interpret it or state the output of a given line — so reading pandas fluently is exactly the examinable skill. Practise saying, in one sentence, what a groupby, .loc filter or merge returns. Chapter 02 · Introduction to Python for Data Science
- Is open data the same as private data?
- No — that is a classic FIT1043 trap. Open data is public, machine-readable data anyone can reuse; the statement 'open data is private data that is machine readable' is false. Machine-readability is a property open data usually has, but the defining feature is that it is openly available, not private. Chapter 03 · Data Sources & Data Wrangling
- Which chart should I use for which question?
- Match the chart to the intent: a histogram for the distribution of one numeric variable, a scatter plot for the relationship between two numeric variables, a boxplot for a five-number summary with outliers, a bar chart to compare categories, a pie chart for parts of a whole, and a line plot for a trend over an ordered axis. The exam often gives a scenario and asks you to pick and justify. Chapter 04 · Data Visualisation & Descriptive Statistics
- What is the difference between supervised and unsupervised learning?
- Supervised learning uses labelled data — every training example has a known target — to learn to predict that target for new inputs; classification (categorical target) and regression (continuous target) are both supervised. Unsupervised learning has no target label and instead finds structure in the data, such as k-means grouping similar points into clusters. Chapter 05 · Data Analysis Theory
- What is the difference between bias and variance?
- Bias is how well a model CAN fit the truth: a simple, low-order model has large bias because it is too rigid to capture a complex pattern. Variance is how much the fitted model changes across different data samples: a complex, high-order model can have large variance because it chases the noise and 'goes wild'. Simple models are large-bias/small-variance; complex models are small-bias/large-variance. Chapter 06 · Regression Analysis
- How do I decide if a problem is classification or regression?
- Look at the target variable. If it is categorical or discrete — spam vs not-spam, pass vs fail, which species — it is classification. If it is a continuous number — price, salary, temperature — it is regression. Predicting a salary is regression because salary is continuous; predicting whether an email is spam is classification. Chapter 07 · Classification & Clustering
- How is R different from Python for this unit?
- They do the same jobs with different syntax. R uses <- for assignment, is heavily vectorised, and reaches for data.frame, lm() and dplyr where Python uses pandas and scipy. The unit teaches both because each is common in industry; the examinable skill is reading either fluently. Week 8 R is also the analysis language for Assignment 2. Chapter 08 · Introduction to R for Data Science
- What are the V's of big data?
- The four core V's are Volume (size of the data), Velocity (the pace at which it arrives and must be processed), Variety (the many different types of data) and Veracity (how accurate and reliable it is). Value is often added as a fifth. The unit frames big data as any V-attribute that challenges a system's capability or a business need — so 'big' is relative to what your system and task can handle. Chapter 09 · Characterising Data & 'Big' Data
- What is the difference between SQL and NoSQL?
- SQL (relational) databases store structured data in tables with a fixed schema and use joins and SQL queries — strong for well-structured, consistent data. NoSQL databases are non-relational and schema-flexible (document, key-value, column-family or graph stores), which suits large-scale, varied or rapidly changing data. The exam point is choosing the right type for the data's structure and scale, not that one is universally better. Chapter 10 · Big Data Processing
- What is the difference between the data lifecycle and the data value chain?
- They are two perspectives on managing data. The data lifecycle is the INTERNAL view — how data is created, stored, used, archived and disposed of within an organisation. The data value chain is the EXTERNAL view — how data flows and creates value across a pipeline such as collect, wrangle, analyse and present. The exam point is being able to differentiate the requirements each perspective imposes. Chapter 11 · Data Governance, Ethics & Privacy
How to study for the exam
Treat FIT1043 as a definitions-and-methods unit and build a one-page recall sheet per week rather than re-reading slides.
Because the final is 15 one-mark MCQs plus 25 two-mark short answers, the highest-yield revision is crisp, justified definitions: the four V's, the k-means five steps, classification vs regression (categorical vs continuous target), sensitivity vs specificity formulas, bias vs variance, the Drew Conway regions, what Hadoop is, the three metadata types, and open data.
Rehearse reading code without writing it — take a pandas groupby/filter line or an R lm() call and say in one sentence what it returns, since both Test 1 and the exam ask exactly that. Do the weekly quizzes and applied-session solutions as formative practice, and keep Weeks 1-4 warm because Test 1 examines them and the final re-examines everything.
For the 50% final, prioritise breadth first: the hurdle rewards being able to start every topic, so make sure you can define and justify across the whole unit before you deepen your weak spots.
When a concept won't click, ask Sia to explain that single idea a different way and to set you a fresh MCQ or short-answer in the exam's style — it teaches the method and checks your reasoning, and it never substitutes for your own graded work. Confirm the exam date, room and rules on the Monash Handbook and Moodle.
Your AI Information Technology tutor for FIT1043
Stuck on a hard FIT1043 question? Sia is AskSia’s AI Information Technology tutor — ask any FIT1043 Introduction to Data Science 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.
