Data wrangling with open
source tools
Tony Hirst
Dept of Communication & Systems
The Open University, UK
Premises
“I take data
from wherever I
can get it”
1
“Appropriate
everything”
2
Conversations
with data
3
Visual
Conversations
with data3
(Accession Plot)
@mediaczar
If a picture’s worth a
thousand words,
maybe it should take
as long to read?
Most learning
analytics won’t be
performed by
learning analytics
researchers
How can we help
people fashion
their own tools to
support data
conversations?
Recipes
site:open.ac.uk
Have a
conversation
with the data…
Ask the right
questions…
xkcd.com/1138
Sometimes a question
makes most sense in
the context of
questions previously
asked and answers
previously received
DATA
USERS
Educators
Learners
Planners
Marketers
Policymakers
Researchers
Press
NGOs
“
D
E
V
E
L
O
P
E
R
S
”
Have
dashboard,
so what?
A tools and
issues
based view
DATA
TOOLS
USERS
PROBLEMS
Example – Google Fusion Tables
Fusion Table
https://www.google.com/fusiontables/DataSource?docid=1VKG7iCbFlsEYJzTuQppf4xoIqq1ABxWTdW6O_7o#rows:id=1
http://is.gd/qhuaoA
Walkthrough
http://blog.ouseful.info/2012/11/16/a-quick-look-at-gcsealevel-certificate-awards-market-share-by-examination-board/
http://is.gd/f9YAbG
DATA
TOOLS
USERS
PROBLEMS
Access/obtain data
Make sense of data
Ask specific questions of data
Communicate in a data-centric way
Load data
Clean data
Merge/enrich data
DATA
Issues
TOOLS
DATA
Other
TOOLS
Issues
TOOLS
“Tool based
programming”
A barrier to access
(for the tool user) is
data format
JSON XMLCSVXLS
TSV
.db
HTML
PDF DOCTXT
GLUE LOGIC(Glue code)
=importHTML(URL, “table”, N)
HTML
QUERYABLE
DATA
Try it…
Example Page
http://en.wikipedia.org/wiki/List_of_colleges_and_universities_in_the_United_States_by_endowment
http://is.gd/7Vbg6n
Google Spreadsheets as a database
Explorer
https://views.scraperwiki.com/run/google_spreadsheet_query/
http://is.gd/jiMJoh
Walkthrough
http://schoolofdata.org/2013/05/24/asking-questions-of-data-garment-factories-data-expedition/
http://is.gd/qJHihu
=importCSV(URL, N)
HTML
INTERACTIVE
DASHBOARD
Google Charts
Google Chart
Visualization API
https://code.google.com/apis/ajax/playground/
http://is.gd/TTHIUh
Google
Visualisation
API
googleVis
(R)
https://developers.facebook.com/
docs/reference/api/examples/
http://is.gd/7cRnvS
A barrier to access
(for the tool user) is
data shape
A barrier to access
(for the tool user) is
data cleanliness
Questions of
identity
The Open University
Open University
OU
Open Uni
Open University, UK
NORMALISATION/RECONCILIATION
Reconciliation to
a canonical name
and/orto a
unique identifier
A stumbling block
(for the data user)
is data enrichment
A stumbling block
(for the data user)
is joining datasets
A stumbling block
(for the data user)
is joining partially
matched data
Rolling your own
interactive data
exploration tools
R Shiny
Apps
ui.R server.R
RCharts
Many chart tools
do the work for
you if the data is
in the right shape
DATA
TOOLS
USERS PROBLEMS
Justask…
ask.SchoolOfData.org
blog.ouseful.info
@psychemedia

Lasi datawrangling

Editor's Notes

  • #25 I am not a journalist, but it seems to me that a large part of your work, and indeed a large part of the work of a scientist or an analyst, is in asking the right questions of a source, and knowing how to frame those questions.The data journalist knows how to ask questions of data.
  • #26 Also – high incidence of crime around police stations (no location, so police station used as default location); Russell Square as a murder hotspot.
  • #27 Another nice example of this, and one used by many advocates of data visualisation, is the famous example of Anscombe’s quartet, for sets of two dimensional data with some interesting properties.
  • #28 For example, many of the “classic” summary statistics for the corresponding columns in these data sets are to all intents and purposes the same.
  • #29 But when we look at the datasets as a set of scatterplots, we see how the data tells very different stories.
  • #126 People learn the skills they need, as they need them.