Defra Science Advisory Council

2016-12-15

Nobel House, London

Neil D. Lawrence

Amazon and University of Sheffield

@lawrennd inverseprobability.com

Mathematical Statistics

• ‘Founded’ by Karl Pearson (1857-1936)

‘Mathematical Data Science’

• ‘Founded’ by ? (?-?)

Today

• Challenges of understanding and interpreting big data today are similar to those that Disraeli faced in with statistics in latter part of the 19th century.

• Data is elusive: it can promise much but deliver little.

Stetson Hats

• What’s the most common hat worn in the west?

The Three ’A’s

• Data Awareness

• Data Availability

• Data Analysis

Data Awareness

• What data you have and where its stored.

• May need to chance conception of what data is and how to obtain it.

• Production lines, smart phones.

• Locked away: manual log books, confidential data, personal data.

• An internal audit (you are ahead here!).

• The key to any successful campaign is a good map.

Data Availability

• How well are the data sources interconnected?

• How well curated are they?

• Curse of Disraeli was associated with unreliable data and unreliable statistics.

• Misrepresentation is worse than absence of data.

• Need an improved sense of data and its value.

Data Analysis

• accumulation of the necessary expertise to digest what the data tells us.

• data requires intepretation, and interpretation requires experience.

• Analysis is a bottleneck due to a skill shortage.

• Ideally, analysis should be carried out by individuals not only skilled in data science but also equipped with the domain knowledge.

“Embodiment Factors”

compute	~10 gigaflops	~ 1000 teraflops?
communicate	~1 gigbit/s	~ 100 bit/s
embodiment (compute/communicate)	10	~ 1013

Evolved Relationship

Societal Effects

• Automated decision making in the computer based on data.

• Need to better understand our own subjective biases better.

• Particularly important where societal interventions are prescribed.

• But what is a societal intervention in the modern era? Much more subtle than before.

Societal Effects

• Shift in dynamic from the direct pathway between human and data to indirect pathway between human and data via the computer

• This change of dynamics gives us the modern and emerging domain of data science

Societal Challenges

1. Paradoxes of the Data Society

2. Quantifying the Value of Data

3. Privacy, Transparency, Fairness, Equality

Challenge 1) Breadth vs Depth Paradox

• Able to quantify to a greater and greater degree the actions of individuals

• But less able to characterize society

• As we measure more, we understand less

What?

• Perhaps greater preponderance of data is making society itself more complex

• Therefore traditional approaches to measurement are failing

• Curate’s egg of a society: it is only ‘measured in parts’

Wood or Tree

• Can either see a wood or a tree.

Examples

• Election polls (UK 2015 elections, EU referendum, US 2016 elections)

• Clinical trials vs personalized medicine: Obtaining statistical power where interventions are subtle. e.g. social media

Also need

• More classical statistics!
  • Like the ‘paperless office’

• A better characterization of human (see later)

• Larger studies (100,000 genome)
  • Combined with complex models: algorithmic challenges

Challenge 2) Quantifying the Value of Data

There’s a sea of data, but most of it is undrinkable

We require data-desalination before it can be consumed!

Value

• How do we measure value in the data economy?
• How do we encourage data workers: curation and management
• Incentivization for sharing and production.
• Quantifying the value in the contribution of each actor.

Credit Allocation

• Direct work on data generates an enormous amount of ‘value’ in the data economy but this is unaccounted in the economy

• Hard because data is difficult to ‘embody’

• Value of shared data: Wellcome Trust 2010 Joint Statement (from the “Foggy Bottom” meeting)

Solutions

• Encourage greater interaction between application domains and data scientists

• Encourage visualization of data

• Adoption of ‘data readiness levels’

• Implications for incentivization schemes

Privacy, Loss of Control and Marginalization

• Society is becoming harder to monitor

• Individual is becoming easier to monitor

• What does it mean if a computer can predict us better than we can ourselves?

Discrimination

• Potential for explicit and implicit discrimination on the basis of race, religion, sexuality, health status

• All prohibited under European law, but can pass unawares, or be implicit

Marginalization

• Credit scoring, insurance, medical treatment
• What if certain sectors of society are under-represented in our aanalysis?
• What if Silicon Valley develops everything for us?

Digital Revolution and Inequality?

Amelioration

• Work to ensure individual retains control of their own data
• We accept privacy in our real lives, need to accept it in our digital

• Control of persona and ability to project

• Need better technological solutions: trust and algorithms.

Societal Challenges

• Paradoxes of the Data Society

• Quantifying the value in the Data

• Privacy, Transparency, Fairness, Equality

Solutions

• Software: TensorFlow, scikit-learn (python), R, Spark.

• Education

• Automating the Data Science Pipeline:
• The Automatic Statistician

• AutoML

Conclusion

• Alone ‘big data’ promises much and delivers little.

• Data needs to be cared for: it needs to be curated and evaluated.

• Thee stages:

1. Awareness
2. Availability
3. Analysis

Conclusion (2)

• Hand waving about big data solutions leads to self-deception.

• The castles we build on our data landscapes must be based on firm foundations.

Thanks!

• twitter: @lawrennd
• blog: http://inverseprobability.com