2. The Urban Data School
• Developed an approach for using large,
complex data sets to teach data literacy skills
in schools
• Used data collected as part of a large ‘smart
city’ project (http://mksmart.org)
• Designed approach in collaboration with
teachers and students in Milton Keynes
3. Key idea 1: Teach in a context that the
student can relate to
• We used data about
‘energy’ – a topic that
students had already
started learning
• We used data that was
from their own town
Solar panel map
Heat map
Data from smart meters (70 MK houses)
4. Key idea 2: follow an inquiry process
Students start by
interpreting outcome of
existing inquiry (guided
inquiry) and this prompts
new round of questions
(open inquiry)
We based all of our lesson plans on
energy research that was being
conducted at the Open University.
5. Key idea 3: work ‘out’ not ‘in’
• The guided part of the
inquiry should always
start from a ‘snapshot’
of the data
• Students interpret one
day of ‘whole house’
energy consumption
from smart meter data.
• Students expand inquiry
to ask questions that
interest them
– look at individual
appliances
– other days/seasons
– other houses…
6. Key idea 4: students collect their own
data
Help students understand
the data by getting them
to collect similar data of
their own
• Students ‘become’ a
smart meter for a week
and record when
energy is used at home
• Students use plasticine
models to learn how
LiDAR data is collected
7. Key idea 5: include creative activities
alongside practical ones
Novel data visualisation App design
8. Summary of Engagement
School Sessions
Wavendon Gate (Primary) Energy tasks
The Academy (Secondary) Energy tasks + app design
Denbigh (S) Energy tasks + app design
Stantonbury (S) Energy tasks
Urban Scholars (various S) App design
UCMK Graduate Fair (S) Energy tasks
Wavendon Gate (P) Energy tasks
+ plasticine modelling
13 SCHOOLS IN TOTAL ≈ 150 students
9. Important Findings
• Students are engaged by local data. Driven to
search for the familiar and pose questions
about things they know
• Teachers need support, too - but they think
the data is an excellent teaching resource
10. Learn to Code for Data Analysis
• Free introduction to coding and data analysis
– Coding for reproducible research
• Futurelearn MOOC: http://tiny.cc/lcda
– 5+ h/week, 4 weeks, in May + Oct each year
• Also on OpenLearn: http://tiny.cc/lcda-ol
– 24/7 but no discussion forums, no support
– used by Space Science MSc module
– materials available under CC-BY-NC-SA license
• 40min session based on week 1 for Year 10
11. Learning
• International open data: WHO, WU, WB, UN.
• Weekly project:
– Start with research questions
– We answer them, introducing necessary concepts
– Interleave reading and doing exercises
– End with data analysis report
• Show and tell: share report on own region
12. First Principles of Instruction
Problem-centred: Base the teaching and learning on interesting and
progressively more complex real-world problems.
1. Activation: Help the learners activate past experience, information or
mental models that can be used to organise the new knowledge.
2. Demonstration: Show the learners the new knowledge, e.g. through
worked examples, preferably with multiple viewpoints.
3. Application: Give learners a sequence of varied problems for them to
apply the new knowledge. Provide feedback and diminishing guidance,
e.g. on how to correct mistakes.
4. Integration: Encourage learners to discuss, reflect on, and publicly
demonstrate their new knowledge or skill, to integrate it into their lives.
More details: http://tiny.cc/fpoi
13.
14. Technology
• Python: easy to learn, used in STEM Faculty
• Pandas: R-like data analysis library
• Jupyter notebooks for exercises and reports
• http://continuum.io/anaconda: computer app
– free for Windows, Mac, Linux
• http://cocalc.com: free web service
– features to distribute, collect, grade assignments
15. Jupyter notebooks
• text editor/formatter + code editor/interpreter
– free professional but easy-to-use software
– data scientists use for reproducible research
– we use in under- and post-graduate courses
• text, code, and outputs (tables, charts, ...)
– handouts with examples and exercises
– assignment reports
– text in Markdown, code in Python, both widely used
• code one line at a time, with immediate feedback
16. Jupyter notebooks
• create and edit notebooks in browser
– students can add own notes and fix typos quickly
• publish read-only version, e.g. Y10 notebook
– export to HTML or PDF and share file
– single-click publish in CoCalc
– put notebook file online, paste URL in
http://nbviewer.jupyter.org
– publish on GitHub (with version control for free!)
17. Demo time!
Plan A: live demo (fingers crossed…)
Plan B: a screencast
Plan C: more slides
18.
19.
20. Problems
• Keeping up with software and site changes
• Provide all data offline because:
– Online (historic) data changes (teaching point!)
– Free CoCalc account doesn’t allow API calls
• File encoding issues
• [](){} and its combinations can be confusing
• Convincing the Excel fans
21. Take away
• Free course for CPD and re-purposing
• Follow First Principles of Instruction
• Problem-driven data analysis is engaging
– It’s interdisciplinary, local, global, personal
22. Take away
• Notebooks are an alternative approach to coding
– code + text (explanation, analysis, study notes)
– Incremental and iterative coding/debugging
• exploratory data analysis without clear research questions
– Require some discipline
• Be aware and prepared for data + software issues
• Coding is a powerful way to improve data literacy
23. Contact
We’d be happy to help you, learn from you, and
work with you.
E-mail: first.lastname@open.ac.uk
Editor's Notes
Data-driven economy, alternative facts in politics, more open data: we need data literate, critical citizens
Part 1: Annika’s work, KS3, I wasn’t involved so hope I can answer some of your questions
Part 2: an approach to teaching data analysis that could be used at KS4+KS5,
Eon data:
1) From single graph when is a) most energy consumed b) least energy consumed
2) Explore graphs of different households: is there a pattern? Write a story theorising the consumption differences within and across households. How could these theories be tested?
Solar panel data: Look first at neighbourhood of 15-20 houses, then at familiar area.
Which of these houses should produce the most solar energy each year (yield)? Is there relation between aspect, pitch, size of panel and yield? What other factors might influence solar energy production? Is one big roof better than many small ones?
heat map: find familiar building, e.g. your school. What is its heat loss? Is it noticeable different from others in surrounding area?
What might explain this difference (if there is one)? What new information would you need to confirm what causes the differences and where might you get this data?
Lidar = light + radar: measure reflected energy/light from laser emission to create 3D representations
TB data from WHO, weather data from WU, GDP + life expectancy from WB, milk+cream trade from UN Comtrade
Activation: could emphasize relation to Excel better
Demonstration: everything towards doing project, few sidetracks; multiple viewpoints hard in short time
Application: exercises based on variations of our code, extension of project; not that varied due to short time; feedback not as effective in massive online setting
Integration: forum discussion, localisation by learners; could have peer-review, sharing is easier with CoCalc
iCMAs far more powerful
No relationship to either company, just mentioning it works for us with many hundreds of students
Used CoCalc and notebook distribution for Y10 session
For those w/o software
Turtology:
Point out interleaved text/explanations and code/exercise
run first 2 cells, one at a time, move turtle window out of way
re-run the individual fd() and left()
Run ping-pong
do the exercise, use tab to auto-complete show/hide
Run rest
Fix wurtle typo, add *To do:* add `wally = Turtle()`
Show Watford notebook if time
UI changes, so where is now command, where do I download WB data? New features like modern notebooks without publishing
Use open data from council to analyse expenditure, make pupils talk to parents
Talk to your science, geography, sociology, language colleagues, e.g. most frequent letters in foreign languages for cryptography
Double reinforcement of concepts in two subjects: learn by teaching to a stupid machine