Data Analysis in the Water Industry:

1

• Data Analysis in the Water Industry
• A Good-Practice Guide with application to SW
• Deborah Gee, Efthalia Anagnostou
• Water Statistics User Group - Scottish Water
• 
  OR54, September 2012

2
Outline of the talk

• Introducing the Business Team
• Project Background
• The Data Analysis Spiral
• Other things included in the Guide
• Key messages

3
Our Business

• provide high quality affordable water
• protect and enhance the environment
• support Scotlands communities and economy

Scottish Water aims to
What the business does

• supply water to 2.4m households 152,000
  businesses
• we manage 97,000km of buried pipes 2,100
  treatment works
• we have 3,700 staff and revenue of 1bn per year

4
Our Team an in-house analytics team

• Vision grow the value of analytics in the water
  industry
• Skill sets statistics, operational research,
  computing asset risk management
• Services develops analytical tools to support
  the business and in particular asset decision
  making.
• Partnerships Universities and Industrial Groups

RISK CONSORTIUM
5
Project Background
shares statistical approaches promote good
practice data analysis across the water industry.
The Water Statistics User Group
Motivation ?
More demand for data driven-decision making in
asset management,
thus a growing need for an in-depth data analysis.
Development approach
3 knowledge elicitation workshops Final draft
and update to WSUG Presentation at the IAM
conference Publish Guide
Jul 2010 - May 2011 Nov 2011
May 2012
6
Part I Data analysis spiral
Part II Basic analysis health checks case
studies
7
Data Analysis Spiral
1
2
Capture Stakeholder Requirements
Gather Business Data
7
3
Conduct Exploratory Data Analysis
Acceptance Test
Increasing Maturity
Increasing acceptance
4
6
Publish Results Identify Opportunities for
Improvement
Develop Analysis
5
Validate Analysis
8
Data Analysis Spiral
1
Capture Stakeholder Requirements
Acceptance Test

• Business need is formulated and confirmed with
  stakeholders.

• The format of the outputs are agreed with the
  stakeholders.

• The appropriate level of uncertainty is agreed
  with stakeholders.

9
Data Analysis Spiral
2
Gather Business Data
3
Conduct Exploratory Data Analysis

• Data is obtained from robust corporate data
  sources or appropriate data collection mechanisms
  are put in place.

• A clear audit trail for the data is established.

• The analyst challenges the data quality and
  develops a good understanding of the data
  composition.

10
Data Analysis Spiral

• An robust methodology is designed, documented
  and applied to the data.

• Underlying assumptions are examined and accuracy
  of the outputs is assessed.

• Pragmatism of the outputs is challenged against
  expert knowledge.

4
Develop Analysis
5
Validate Analysis
11
Data Analysis Spiral

• Outputs from the current iteration of the spiral
  are finalised and released to the stakeholders.

• Documentation is prepared for technical and
  non-technical audiences, alongside training
  material.

• Recommendations for improvement are identified
  and the maturity of the analysis is assessed.

6
Publish Results Identify Opportunities for
Improvement
12
Data Analysis Spiral
Capture Stakeholder Requirements
7
Acceptance Test

• Stakeholders provide detailed feedback to the
  analyst.

• A further iteration of the Data Analysis Spiral
  is initiated if the stakeholder is not satisfied.

13
Other things included in the guide
real-world

• examples of best-practice for each step of the
  Spiral.
• describe potential consequences when
  best-practice is not applied

Case Studies
? the analyst provides the stakeholder with
analysis proposal ? the data can be audited ?
documentation is version controlled
Analysis Health Checks
a simple to-do list
14
What are the key messages?
The growing need for robust data analysis and
data management is reflected across all asset
management sectors.
Using the good practice guide, analysts can
demonstrate transparency, consistency and
quality in their analysis.
Within SW the guide ? is a benchmark for
assessing data analysis.
? creates a standard process for data
analysis
which meets the requirements for ISO9001.
? inform
stakeholders of what good analysis is.
15

• If you would like a copy of the guide please
  contact us
• Deborah.gee_at_scottishwater.co.uk
• Efthalia.anagnostou_at_scottishwater.co.uk
• Thank you