Samos Summit 2015 - Policy Compass

Grant Agreement

612133

Understanding the policy impacts

using Open Data: The Policy Compass approach

Dr. Yury Glikman

Fraunhofer FOKUS

30 June 2015 Samos Summit 2015

http://policycompass.eu

http://policycompass.eu/

• Project start: 1 October 2013

• Duration: 36 months

• Project Coordinator: Fraunhofer FOKUS (Germany)

• Consortium: 7 organizations from 5 different countries

Policy Compass Consortium

30 June 2015 Samos Summit 2015 2

Policy Compass in 3 words

Learning from Data


Traditional Open Data Platforms

• Metadata registry

• Data storage (optional)

• Basic visualisation of single data files (optional)

• Different data file formats:

• Machine processing friendly: csv, well structured excel files

• Human reader friendly: pdf

• Different data structures

• Metadata and data in local language only (the usual case)


The Problem

• Difficulty of objectively assessing

the impact of government policies

• Political discussions about policies

impacts are highly emotional and

often are not based on any factual

data

• Poor use of available open data

resources in the internet

• Different perception by people

about a suitable metric for

measuring progress

The Opportunity

The Problem and the Opportunity

The Opportunity: learning from data


ONLINE DELIBERATION

AND ARGUMENT MAPPING

BUILDING

CAUSAL MODELS

DEFINING OWN

PROGRESS METRICS

5

Policy Compass key facts

• Policy Compass is an Online Web platform

• Policy compass works with time series data

• Users can define own metrics by providing a formula for its

computation

• Policy Compass draws line charts for selected by the user data or

metrics

• User can annotate charts by the policy changing events information

• User can define causal models explaining behavior of the chart

curves and simulate future behaviour

• Users can discuss the causal models, charts, etc

• Discussion results can be formalized as an argumentation graph

• Policy Compass is not an A.I. tool



• Policy Compass is a Online Web platform



computation


metrics












computation


metrics












computation


metrics












computation


metrics












computation


metrics












computation


metrics












computation


metrics












computation


metrics












computation


metrics








Policy Compass Process


The Policy Compass Concept Components

MetricsDefine own data based metrics

VisualisationsVisualise and compare metrics

Historical eventsAnnotate metrics visualisations by

policy relevant events

ArgumentationTransforms statements and

opinions into structured argument

maps

DeliberationDiscuss metrics and policy models

Fuzzy Cognitive MapsModel policy impacts and

interrelations


Find or upload time series data

• Policy Compass has already some data uploaded by other users

• If interesting for you data is missing, try to look at Open Data or

Statistical portals, like Eurostat, for example

• Provide for uploaded data infomation about ist name, origin and other

attributes

• Specify for which indicator(s) are these data. Think about indicators as

about a group to which these data belong.


Find metrics of your interest or define your own

• There are already some

metrics defined by other

users.

• If you want to define you

own one you can do it

by providing a formula

for it.

• In your metric formula

you can use available in

the platform indicators

(you can add new if

necessary), other

metrics and standard

mathematic operations.


Visualise selected indicators or metrics as timeline charts


Look for explanations of metrics or indicators change


• Annotate the chart by historical events, which could be a reason for

the change

Historical

events

21

Define a causal model explaining your theory


• Define causal model

• Assign weights

manually or ask Policy

Compass to suggest

them

• Perfom simulation to

predict further

development of the

indicators and metrics

22

Discuss your findings with the community

• Discuss your findings with the community

• Embed any of the listed above steps in your blog or web site

• Formalise the discussion outcomes as an argumentation graph


Correlations can be interesting , but useless


Correlations can be interesting , but useless


www.PolicyCompass.eu

www.twitter.com/PolicyCompassEU

www.facebook.com/PolicyCompass

PolicyCompass

Thank you!

Grant Agreement

612133

Yury Glikman

[email protected]

Policy Compass Terminology

• Property: A binary predicate denoting an attribute of an individual.

‒ Example: prosperity, health, criminality

• Indicator: A property correlated with some other property.

‒ Example: average income, number of deaths, number of drug law offences

• Dataset: A table for some property of a set of individuals over a time

period, where each row is a time series and each column is a time

point, e.g. year, and the cells contain the value of the property for a

particular individual and time point.

‒ Example: Average income in UK in the years 2000-2014; number of deaths in Spain

in 2001; number of drug law offences for European countries in 2000-2013.

• Metric: A measure of some property, a mathematical function

(formula) from the set of individuals to data values, defined using

indicators.

‒ Example: Deaths rate = number of deaths / country population


Samos Summit 2015 - Policy Compass

Science