CSCI 3141 · R and Python · Dalhousie University
Taught by Prof. Gabriel Spadon
Welcome to the public materials for CSCI 3141. This course follows a data science project from the first question to an analysis that someone else can understand, check, and reproduce. You will work with data, make decisions about its quality, explore patterns, build and evaluate models, and communicate what the evidence supports.
The repository is a place to prepare for class, revisit a difficult example, and return to methods after the course ends. Independent readers are welcome to use the released materials too. You can read and download them without a GitHub account.
This guide describes the whole course, including material awaiting publication. The download list identifies the documents you can open now. For an enrolled student, the syllabus for your offering is the authority on dates, meeting details, assessment requirements, and policies. Check that it matches the offering identified by your instructor.
- Start here
- Available materials
- What you will learn
- The course week by week
- How to work through a week
- R, Python, and your working environment
- Labs, assignments, and examinations
- Your data science portfolio
- Collaboration and permitted tools
- Readings and further study
- Questions, access, and support
- Sharing and attribution
Read the syllabus first, then open the first class's slides and lecture notes together. The slides give you the main ideas, diagrams, and examples. The notes develop the reasoning and provide worked steps you can return to at your own pace.
Choose either R or Python and prepare one local or cloud environment. Try a small example from the lecture, save your work, and run it again from the beginning. This gives you an early chance to resolve access or setup problems.
If you are enrolled, check your access to the course Microsoft Teams space and follow the syllabus instructions for setting up your private assessment repository. A GitHub account is needed for that portfolio workflow. Bring a laptop to practical meetings, or contact the instructor early if computer access is a barrier.
Students from any major are welcome. The syllabus lists no formal prerequisite course, but expects preparation in calculus, statistics, basic computing, and programming. You are not expected to arrive knowing every package or model used later in the course. Guided examples introduce the tools before assessed use.
Released teaching documents are provided as PDFs. GitHub can preview them in your browser, and you can download copies for annotation or offline reading.
- Course syllabus
- Week 1, Class A, Course Orientation and the Data Science Process
- Week 1, Class B, Summarizing Data
- Week 2, Class A, Data Preprocessing Foundations
- Week 2, Class B, Cleaning and Transforming Data
- Week 3, Class A, Preprocessing and Visualizing Data
- Week 3, Class B, Visualizing Data
- Week 4, Class A, Manipulating and Visualizing Data
- Week 4, Class B, Reproducible Reporting
The complete teaching sets for Weeks 1–4 are available above, each with both classes and its integrated lab. Assignment 01 is also available. Weeks 5–12 and Assignments 02–04 are not yet published here. The course plan below describes those later topics so you can see how they fit together. Download links will be added as documents are released.
The folders follow the teaching sequence. Within each week-XX folder,
Class A and Class B are the first and second class of that teaching week.
Assessment is the location for released lab and assignment handouts.
extras holds the notebooks run in class, where a week has them.
Course-wide documents are under assets/. A folder may exist before its
documents are available; an empty folder does not add a preparation or submission
requirement.
The course starts with a table of observations and the question that brought you to it. What does a row represent? Which variables were measured, and in what units? What is missing? Before a model can be useful, you need a defensible answer to those questions.
From that foundation, you will learn to import, clean, transform, and document data; choose summaries and graphics that fit the variables; and explain the patterns you find. Later, predictive methods use observed outcomes to learn how to estimate an outcome for another case. Descriptive methods examine structure, similarity, groups, and unusual observations without requiring a target outcome.
By the end of the course, you should be able to carry an analysis through these decisions and explain why you made them. That includes comparing models using appropriate evidence, acknowledging uncertainty and limitations, and preparing a reproducible report in R Markdown or a Python notebook.
These steps are connected. An unexpected plot can reveal a cleaning problem; an evaluation result can send you back to your choice of variables. The course returns to the data science process as new methods become available, so the methods remain connected to the question they are meant to answer.
The following learning path follows the syllabus. Week numbers refer to teaching weeks, so consult the syllabus for calendar dates, breaks, releases, and deadlines. Pace and examples may be adjusted through course announcements.
- Week 1 · Course orientation and summarizing data. Class A introduces the data science process and the course workflow. Class B develops observational units, variable types, descriptive summaries, and plots. Lab 01 establishes a descriptive baseline and the private submission process. Models shown during orientation are previews; this lab does not require model fitting or significance tests.
- Week 2 · Data preprocessing. Class A establishes the foundations of data preparation. Class B develops cleaning, transformation, and feature engineering, including how preparation decisions change the table used for analysis. The practical work builds on Week 1's understanding of variables and summaries.
- Week 3 · Preprocessing and visualization. The classes connect prepared data to visual exploration, then examine visualization in more depth. You practise choosing a display for the question and reading what its scales, groups, and distributions actually show. Assignment 1 is released in this part of the course.
- Week 4 · Manipulation and reproducible reporting. Class A brings together data manipulation and visualization in R and Python. Class B develops reporting in R Markdown and Python notebooks. The analysis, figures, and explanation become a document that can be rerun and checked.
- Week 5 · Predictive tasks and metrics. Classification and regression introduce two different kinds of prediction problem. The classes examine their corresponding performance measures, building the vocabulary needed to judge a model's results.
- Week 6 · Predictive methods. The data science process is revisited through predictive metrics, followed by k-nearest neighbours, linear regression, and linear discriminant analysis. Attention moves from describing a dataset to fitting a method and interpreting its predictions.
- Week 7 · Midterm and support vector machines. The first meeting is the midterm examination. The second introduces support vector machines and continues the predictive modelling sequence. Assignment 2 is released in this part of the course.
- Week 8 · Decision trees. The classes introduce decision trees and develop tree-based models and pruning. You examine how a tree partitions data and how its complexity affects the result.
- Week 9 · Performance estimation. Both classes focus on how model performance is estimated. The central question is how much confidence an evaluation gives you about performance beyond the data used to fit a model. Assignment 3 brings the predictive workflow together.
- Week 10 · Model complexity and descriptive analytics. Bias, variance, overfitting, and regularization lead into descriptive analytics, similarity, and distance. The focus broadens from predicting a specified outcome to studying the structure within a dataset. Assignment 4 is released here.
- Week 11 · Clustering. Distance and preparation choices lead into clustering algorithms and their evaluation. You examine how the definition of similarity affects the groups an analysis produces.
- Week 12 · Outliers and workflow synthesis. Outlier detection is followed by cumulative review. The final class connects preparation, visualization, predictive and descriptive modelling, evaluation, and reporting into the complete workflow.
Some lectures include optional research examples or extensions. Their presence in a slide deck does not make them a prerequisite for that week's assessed work. Use the lab or assignment handout to identify the methods and evidence required.
Before a class, read its opening goals and review the ideas it carries forward from the preceding meeting. During class, use the slides to follow the sequence of questions, examples, and diagrams. Mark the points you want to revisit in the notes rather than trying to transcribe every slide.
Afterwards, work through the corresponding explanation and examples in the notes. Run the code yourself and compare the result with the stated interpretation. If the outputs differ, first check the input data, selected rows, variable types, and preparation steps. Keep a short record of the question you were trying to answer and the decision each step supports.
Each week's second class begins an integrated lab. Use it to apply the methods introduced so far, ask questions, and develop your own analysis. The lab handout sets the approved data, required work, deliverables, and rubric. Keep it beside the slides and notes while you work.
As reports become more substantial, practise restarting the R session or Python kernel and running the analysis from beginning to end. A saved output should be traceable to the code and data in the submitted version. Explain a result in words as well as displaying it, including its units, relevant sample size, and any limitation needed to interpret it fairly.
R is the primary teaching language, with full Python parity in materials, demonstrations, practical work, and assessments. Choose one language for your work. Both follow the same requirements and grading standards; you do not need to submit an analysis twice in two languages.
You can use a local R or Python environment, or one of the cloud options listed in the syllabus.
- Posit Cloud provides browser-based RStudio and Jupyter projects for R or Python work.
- Kaggle Notebooks supports R and Python notebooks.
- Google Colab provides Python notebooks in the browser.
Another suitable free cloud notebook service is also acceptable. No paid external service or public generative AI service is required. Cloud plans and resource limits can change; consult the provider's guidance when choosing an environment. Keep copies of the notebooks and supporting files in your private assessment repository so the work is preserved beyond a running session.
Use the software and dataset guidance supplied with the relevant class or activity. Record the data source and the steps needed to run your analysis. Check a cloud notebook's sharing settings before putting assessed work in it. Permission to use an editor or notebook service does not authorize its generative AI features for assessed work; the rules below still apply.
The numerical grading structure in the published syllabus is shown below. The syllabus contains the full assessment and passing requirements, approved aids, late-work rules, and absence and accommodation procedures. Consult it alongside the instructions for each assessment.
| Graded component | Course weight |
|---|---|
| Integrated labs | 15% |
| Four individual assignments | 45% |
| Midterm examination | 15% |
| Cumulative final examination | 25% |
The best 10 of 12 individual lab results count toward the lab component. Labs are due seven calendar days after the integrated session, before the next usual Class B start time. The current syllabus specifies 16:05 Halifax time, including weeks without a scheduled meeting. Assignment release and due dates follow the syllabus; they do not use the weekly lab deadline.
Labs provide regular practice with the current methods. The four assignments ask you to assemble a more complete analysis, with increasing responsibility for preparation, methodological choices, interpretation, and reproducibility.
| Assignment | Primary focus | Course weight |
|---|---|---|
| 1 | Import and basic exploration | 7.5% |
| 2 | Preprocess and explore data | 7.5% |
| 3 | Predictive modelling solution | 15% |
| 4 | Descriptive modelling solution | 15% |
Use the released handout for the dataset, required files, format, rubric, and authorized tools. Assessment handouts state what to demonstrate; students must write and justify their own analysis. Worked teaching examples are in the lecture materials. The submitted files must open and reproduce the analysis. Submission dates belong to the syllabus and the instructions for your offering; this course guide does not set a separate calendar.
The midterm is held in class, and the Registrar schedules the cumulative final examination. Follow the examination's approved-aids instructions and the syllabus when preparing for either.
Your labs and assignments become a portfolio of analyses you can discuss in class and revisit later. A reader should be able to find the question, identify the data, follow the analysis, and understand the conclusion without needing you to explain every file in person.
The syllabus uses these locations in your own assessment repository.
labs/
lab-01/
lab-02/
...
assignments/
assignment-01/
assignment-02/
...
Follow each handout for the files inside its folder. These directories belong
to your own work repository. Each one carries its own README.md, which is
where you record the version to be graded; course-repository issues and pull
requests are not assessment submissions.
Work in a private GitHub repository with instructor access. Commit regularly as you work and push those commits to the private repository. Keep code, reports, figures, outputs, cloud notebooks, and links that reveal answers private before the deadline and until sharing is opened. A branch in a public repository is public too.
First commit the completed report, source, and supporting evidence. Record that
commit's full SHA and commit URL in the assessment directory's README.md,
then commit the README update separately. A commit cannot contain its own SHA.
Push both commits to the default branch of your private repository by the
deadline. The marker reads that README and assesses the snapshot it names.
A branch link or movable tag is insufficient because its destination can
change. Keep the submitted commit reachable.
Preserve recorded assessment history permanently. Do not amend, rebase, squash, delete, replace, or force-push those commits, move submitted tags, or fabricate or backdate progress. Make corrections in new commits. If you accidentally commit private information, contact the instructor for an approved correction.
Git records a version's contents; its author dates do not prove when you submitted the work, and Git alone does not establish authorship or prevent history changes. The pushed README record, its recorded SHA, and the instructor-retained snapshot provide the independent assessment reference.
After the submission deadline and when the instructor opens sharing, publish the same submitted commit and its history in your public portfolio. First inspect every file in that commit and its reachable ancestors. Every assessment present there must have passed its deadline and been released for sharing by the instructor. Otherwise keep the repository private and use the submitted private snapshot until publication is authorized. Do not rewrite history to remove unreleased work. Preserve its SHA when merging from the private work repository. Present that version in class, and label any later improvements separately so a reader can distinguish the submission from subsequent work.
Grades and feedback remain private. Where accommodation, safety, privacy, copyright, or licensing requires a private portfolio, the syllabus provides an equivalent private repository option with the same evidence and rubric.
Discussing concepts is encouraged. Assessed work must demonstrate your own understanding. Do not share answers, code, analysis files, answer-producing prompts, or solution steps unless written instructions authorize collaboration. Ask about your own unreleased solution privately, through Microsoft Teams or e-mail to the instructor, rather than in a shared space where classmates read it.
The syllabus permits generative AI for private study planning and practice, provided you check outputs against course sources and protect private or copyrighted information. It also permits editing a student-written ungraded note when the intellectual work is already the student's.
Generated code, prose, analysis, and figures cannot be submitted for integrated lab credit. Generative AI is prohibited for assignments unless the instructions expressly authorize a named use, and it is prohibited in examinations. An authorized assignment use requires disclosure of the tool, purpose, interaction summary, and changes made. Silence in the instructions is not permission. Do not upload non-public course or student material to a generative AI service.
See the syllabus for the complete academic integrity policy and the scope of any authorized use in an assessment's instructions.
There is no required paid textbook. The released slides, lecture notes, and practical handouts are the starting point for learning the course material. Use the reading and documentation links supplied with the corresponding class when you need another explanation or more detail about a method.
Optional further reading is identified in the teaching materials. Those texts and research examples extend the discussion; they do not create additional assessment requirements. Start with the class explanation, then pursue the references that address the question you are working on. Check Dalhousie Libraries for access before obtaining a book.
When studying independently, follow the teaching order and run the examples in your chosen language. For an early analysis, concentrate on the variables, summaries, plots, and preparation steps introduced so far. Return to that analysis as modelling and evaluation methods become available to you.
Use the course Microsoft Teams space for class-wide questions about concepts, examples, or corrections. Include the document title and page or slide number. For a code question, explain what you expected, what happened, and the smallest example that shows the issue, while keeping assessed solutions private.
For individual concerns, contact Prof. Gabriel Spadon at spadon@dal.ca from your Dalhousie email. Office hours are by appointment through Microsoft Teams. Independent readers can also email questions or corrections.
Contact the instructor early if computer access, a document format, or another barrier affects participation. The syllabus links the Student Accessibility Centre and other university support services and explains the processes for absences and accommodations. Personal circumstances, grades, and feedback should be discussed privately.
Enrolled students should check the course repository and their Dalhousie email regularly, and follow the announcement channels identified for their offering. Meeting locations, office-hour arrangements, and calendar dates should always be checked against that offering's syllabus and announcements.
Unless the instructor explicitly marks material private or restricted, released course materials may be shared with credit to the author and CSCI 3141. For example, credit a reused course slide to “Gabriel Spadon, CSCI 3141, Foundations of Data Science, Dalhousie University” and include a link to the source document.
Preserve source links, third-party attributions, and applicable licence terms for figures, datasets, code, and quotations. Copyright remains with the respective owners. Permission to share course materials does not authorize publishing another student's work, restricted data, grades, feedback, or instructor-marked private information. Your own assessed work follows the portfolio release process above.