Applying the data science process to a scenario

Name:

Date:

Class:

Quick Questions

Shade in the A or B circle for each question.

1. A Year 9 class wants to know if students sleep enough on school nights. Which stage comes FIRST in the data science process?

A Define the objective (the research question)

B Draw a chart of the results

2. Survey scenario: students have finished filling in an online survey about screen time. What is the next stage?

A Validate and clean the responses

B Report insights to the principal

3. Survey scenario: 12 responses say the student is 150 years old. What should you do?

A Flag them as invalid during cleaning and decide whether to fix or remove them

B Keep them because every response counts equally

4. Sensor scenario: you want to know if the school oval is too hot at lunch. Which is the best way to ACQUIRE the data?

A Ask one friend to guess the temperature

B Log readings from a temperature sensor at regular intervals

5. Sensor scenario: the temperature log shows -999 for several readings. Which action is correct?

A Include them in the average because the sensor recorded them

B Treat them as sensor faults, then clean or remove them before analysing

Match the Pairs

Draw a line connecting each item on the left to its match on the right.

Left column		Right column
1Stage 1: Define the objective		AThe stages are in the wrong order. The objective must be defined before data is collected or analysed.
2Stage 3: Validate and clean		BWrite a clear research question first, such as 'Do Year 9 students who sleep more score higher in tests?', so you know what data to collect.
3Repository scenario: acquire data by downloading a CSV from an open data portal, then check its licence and filter it to the rows you need.		CIF missing_readings > 30% THEN go back to Stage 2: Acquire
4Going back a stage: validation shows many sensor readings are missing, so return to the acquire stage and collect more data before analysing.		DStage 2: Acquire -> download rainfall.csv, filter Year = 2024
5Survey scenario: analyse the results first, then write the research question after seeing the answers.		ECheck each reading or response against rules and fix errors, for example removing a sensor reading of -999 or a survey age of 250.
6Sensor scenario: the temperature log has errors, but we skip cleaning and report the average anyway.		FSkipping validation and cleaning means faulty readings distort the average, so the insights reported are unreliable.

Definitions

data science process

Fill in the Blanks

Complete each line of code by filling in the missing parts.

______("Do teens sleep enough?") data ← acquire("survey")

data ← acquire("temperature sensor") data ← ________(data) analyse(data)

data ← acquire("open data portal") data ← validate(data) data ← _____(data)

IF quality = "poor" THEN GO BACK TO _______ ENDIF

insights ← analyse(data) ______(insights)

Spot the Error

Shade in the letter of the line that contains an error.

Question 1

A report: how much do we sleep?

B acquire: survey 30 students

C validate: ages must be 13-16

D report: share the insight

Question 2

A define: is the room too hot?

B acquire: guess the temps

C validate: drop faulty -99s

D analyse: find the average

Question 3

A define: do cities get hotter?

B analyse: download open CSV

C clean: remove duplicate rows

D analyse: compare yearly means

Question 4

A if too many errors: analyse

B acquire: collect data again

C validate: check the new data

Question 5

A define: best canteen food?

B acquire: survey 50 students

C clean: merge Pizza and pizza

D analyse: ignore the answers

E report: show a bar chart

Answer Key

Applying the data science process to a scenario

Section A: Quick Questions

A — Define the objective (the research question)

A — Validate and clean the responses

A — Flag them as invalid during cleaning and decide whether to fix or remove them

B — Log readings from a temperature sensor at regular intervals

B — Treat them as sensor faults, then clean or remove them before analysing

Section B: Match the Pairs

Stage 1: Define the objective → Write a clear research question first, such as 'Do Year 9 students who sleep more score higher in tests?', so you know what data to collect.

Stage 3: Validate and clean → Check each reading or response against rules and fix errors, for example removing a sensor reading of -999 or a survey age of 250.

Repository scenario: acquire data by downloading a CSV from an open data portal, then check its licence and filter it to the rows you need. → Stage 2: Acquire -> download rainfall.csv, filter Year = 2024

Going back a stage: validation shows many sensor readings are missing, so return to the acquire stage and collect more data before analysing. → IF missing_readings > 30% THEN go back to Stage 2: Acquire

Survey scenario: analyse the results first, then write the research question after seeing the answers. → The stages are in the wrong order. The objective must be defined before data is collected or analysed.

Sensor scenario: the temperature log has errors, but we skip cleaning and report the average anyway. → Skipping validation and cleaning means faulty readings distort the average, so the insights reported are unreliable.

Section C: Definitions

data science process — Work through a scenario in stages: define the objective, acquire data, validate and clean it, analyse it, report insights, and go back a stage if the data is poor.

Section D: Fill in the Blanks

define("Do teens sleep enough?")
data ← acquire("survey")

data ← acquire("temperature sensor")
data ← validate(data)
analyse(data)

data ← acquire("open data portal")
data ← validate(data)
data ← clean(data)

IF quality = "poor" THEN
  GO BACK TO acquire
ENDIF

insights ← analyse(data)
report(insights)

Section E: Spot the Error

A — The process starts by defining the objective, so line 1 must be the define stage. Reporting insights is the last stage. It should read: define: how much do we sleep?

B — In a sensor scenario the acquire stage collects real readings from the sensor, not guesses. It should read: acquire: log temp every 10 min

B — Downloading data from a repository is the acquire stage, not the analyse stage. Analysis happens after the data is acquired and cleaned. It should read: acquire: download open CSV

A — When data is poor you must not analyse it. You go back a stage, such as acquiring better data. It should read: if too many errors: go back

D — The analyse stage works with the cleaned data to find patterns, such as counting each answer. Ignoring the answers would make the survey pointless. It should read: analyse: count each answer