Learning outcomes (RA)¶

When you start a new video game, the first thing you look for is the quest log. What do I have to do? What counts as "done"? How much is each quest worth? Nobody enjoys finding out the rules after the boss fight.

This module has a quest log too. It is short: three quests. Every class, every practical and every mark you get this year hangs from one of them. Read this page once and you will always know why we are doing what we are doing — and what you are being graded on.

What is a learning outcome?¶

A learning outcome is one thing you must be able to do by the end of the module. In Catalan it is a resultat d'aprenentatge, and everybody shortens it to RA.

Notice the word do. An RA is not a list of topics to memorise. It is a skill: "extracts data", "cleans data", "visualises data". The verbs are the point.

An RA on its own is too big to grade. "Visualises data" — how well? With what? So each RA is split into smaller, checkable statements:

An assessment criterion is one concrete check inside an RA. If the RA is the quest, the criteria are its objectives: the list you tick off to complete it.

And you do not train for a quest all at once. The course is cut into blocks of classes:

A work unit (unitat de treball, UT) is a block of sessions on one topic. Each UT trains one RA.

The map: three outcomes, four units¶

The module follows the data lifecycle: the path data takes from the place it lives to the screen where somebody reads it. In class we compare it to a crafting video game: first you gather materials, then you refine them, then you show the result.

flowchart LR
  subgraph EV1["First evaluation · 22 h"]
    RA1["<b>RA1 · Extract</b><br>(farm and loot)<br>UT1 · 10 h"]
    RA2["<b>RA2 · Clean and transform</b><br>(smelt, sort, craft)<br>UT2 · 12 h"]
  end
  subgraph EV2["Second evaluation · 18 h"]
    RA3["<b>RA3 · Visualise</b><br>(the HUD and the minimap)<br>UT3 · 8 h + UT4 · 10 h"]
  end
  RA1 -- "raw data" --> RA2
  RA2 -- "clean data" --> RA3

The module runs for about 40 hours: 2 hours a week, from 21 September 2026 to 1 March 2027.

RA1 · Extract data from different storage platforms¶

Official wording: Extreu dades de diverses plataformes d'emmagatzematge de dades.

Data never arrives on your desk ready to use. It sits in a database behind a password, in a file somebody exported, or behind a web address that answers in a format you have never seen. RA1 is the "farm and loot" stage: knowing where the materials are and how to bring them home.

UT1 ends with an assessed practical that checks a), b) and c) together.

RA2 · Clean and transform the data for its classification¶

Official wording: Depura i transforma les dades per a la seva classificació.

Real data is dirty. Dates written three different ways. Prices missing. The same game listed twice. A number stored as text. RA2 is the "smelt and sort" stage: you turn raw ore into something you can build with. It is the least glamorous outcome and the one professionals spend the most time on.

UT2 ends with an assessed practical that checks a), b) and c) together, and closes the first evaluation.

RA3 · Visualise the data, representing it graphically¶

Official wording: Visualitza les dades, representant-les gràficament.

A table of 26,000 rows answers no question at a glance. A good chart answers it in two seconds. RA3 is the HUD and the minimap of the game: the health bar, the map, the score — everything that matters, readable without thinking. It is the outcome with the most weight (40%) and the only one that spans two units.

RA3 is graded in two parts:

How your mark is calculated¶

Three steps, from the smallest piece to the final number.

Step 1 — the mark of each RA. Inside every RA, your work is split in two:

Step 2 — pass every RA. You need at least a 5 in each RA. This is the rule that surprises people:

You cannot pass the module with a failed RA. A 9 in RA3 does not rescue a 4 in RA1.

Step 3 — the final mark. Once all three are passed, they are combined by weight: RA1 30%, RA2 30%, RA3 40%.

If you fail an RA¶

A recovered RA is capped at 5. Recovery gets you through; it does not improve your average. The cheapest strategy is to pass each RA the first time.

Rules that cost marks¶

Presentation and spelling count too. The only AI assistant authorised in this course is cs50.ai — anything else needs the teacher's explicit permission first.

Key takeaways¶

• The module has three learning outcomes (RA): extract (RA1), clean and transform (RA2), visualise (RA3). They follow the data lifecycle.
• Each RA has three assessment criteria — a), b), c) — and those nine checks are what every practical and exam measures.
• Four work units (UT) train the three RAs: UT1 → RA1, UT2 → RA2, UT3 and UT4 → RA3.
• Each RA mark is 70% tests + 30% class work; the final mark weighs the RAs 30% / 30% / 40%.
• You must pass every RA with at least a 5. A high average does not cover a failed one.
• A recovered RA is capped at 5, and late work is capped at 5 (0 after 48 hours).

Quick check¶

Now you know the quests. The materials to complete them — every dataset we use in class — are on the Resources page.