Developing insight from commercial data to support #Census2022

Transcript

• 1. Developing insight from commercial data to support #Census2022 BSPS Conference – September 2014 Andy Newing a.newing@soton.ac.uk Ben Anderson b.anderson@soton.ac.uk (@dataknut) Sustainable Energy Research Group
• 2. Census2022: Extracting value from near real time …….. RGS Aug 2014 Overview 2
• 3. Census2022: Extracting value from near real time …….. RGS Aug 2014 What we are trying to do: Census2022  UK Census 2011/2021 evolution  Timeliness & cost  Challenges  Finding new ways to deliver the Census – ‘Census-like’  Opportunities  New kinds of data  New kinds of social indicators - ‘Census-plus’  More frequently 3
• 4. Census2022: Extracting value from near real time …….. RGS Aug 2014 Smart metering • Universal mandate • Quasi-real time • High temporal resolution • Geo-coded • Reveals actual behaviors • Near 100% coverage 4
• 5. Census2022: Extracting value from near real time …….. RGS Aug 2014 Smart metering 5 Smart Gas Meter Smart Electricity Meter WAN DCC Utilities and Authorised Third Parties 3 In Home Display(s) Utility World managed by the utility Consumer World managed by the consumer 2 1 Bills etc. SMHAN Smart Meter Home Area Network Comms Hub Consumer Gateway(s) •Appliances •Consumer HAN •Internet •Services •Future See also http://www.gov.uk/government/policies/helping-households-to-cut-their-energy-bills/supporting-pages/smart-meters
• 6. Census2022: Extracting value from near real time …….. RGS Aug 2014 Electricity Load profiles • Household composition & characteristics • Ownership and use of appliances • Habits and routines 6
• 7. Census2022: Extracting value from near real time …….. RGS Aug 2014 Generating area based household statistics and indicators Household Load Profiles Infer household characteristics Aggregate to small area geographies
• 8. Census2022: Extracting value from near real time …….. RGS Aug 2014 Smart meter-like dataset
• 9. Census2022: Extracting value from near real time …….. RGS Aug 2014 UoS Energy Monitoring Study (UoS-E) 9  Smart meter-like household dataset  n=180  Repeated surveys:  characteristics, behaviors and attitudes  1 second level power import  Sample: October 2011  ~ 500m records (1 second)  Cleaned & checked  Aggregated (mean power)  ~ 250,000 records (half hourly)
• 10. Census2022: Extracting value from near real time …….. RGS Aug 2014 10 Descriptive Analysis 1-2 persons vs 3+ Midweek: No children vs 1-2 vs 3+ Midweek: Respondent in employment vs not
• 11. Census2022: Extracting value from near real time …….. RGS Aug 2014 Analytically: Load profile indicators 11
• 12. Census2022: Extracting value from near real time …….. RGS Aug 2014 Evening consumption factor (ECF) Midweek (Tuesday – Thursday) Ratio of mean 30 minute evening peak power import (4pm – 8pm) to off peak power import Ψ note: n= 5 12 ECF All households Employed Not in active employment All households 2.13 1.64 No Children 2.21 2.54 2.09 With Children 2.31 2.29 1.30Ψ
• 13. Census2022: Extracting value from near real time …….. RGS Aug 2014 Predicting household characteristics
• 14. Census2022: Extracting value from near real time …….. RGS Aug 2014 Inferring household characteristics Presumption of availability via administrative sources • Exploratory analysis suggests clear links but poor explanatory/predictive power • However … improvements in model fit are encouraging given limitations of the dataset
• 15. Census2022: Extracting value from near real time …….. RGS Aug 2014 Inferring household characteristics using classification • Classification / cluster Profile analysis Indicators • Applied within electricity sector to cluster households based on their consumption • Our interest is in underlying characteristics • Partitional clustering technique (k-means) Consumption driven clusters Link to characteristics of interest
• 16. Census2022: Extracting value from near real time …….. RGS Aug 2014
• 17. Census2022: Extracting value from near real time …….. RGS Aug 2014 Generating area based statistics
• 18. Census2022: Extracting value from near real time …….. RGS Aug 2014 In an ideal world… 18 This project
• 19. Census2022: Extracting value from near real time …….. RGS Aug 2014 Realising the ‘added value’ from domestic smart metering 19 Smart Gas Meter Smart Electricity Meter WAN DCC Utilities and Authorised Third Parties 3 In Home Display(s) Utility World managed by the utility Consumer World managed by the consumer 2 1 Bills etc. SMHAN Smart Meter Home Area Network Comms Hub Consumer Gateway(s) •Appliances •Consumer HAN •Internet •Services •Future See also http://www.gov.uk/government/policies/helping-households-to-cut-their-energy-bills/supporting-pages/smart-meters
• 20. Census2022: Extracting value from near real time …….. RGS Aug 2014 Thank you  http://www.energy.soton.ac.uk/tag/census2022/  Ben Anderson b.anderson@soton.ac.uk (@dataknut)  Andy Newing a.newing@soton.ac.uk 20