Sofia Pinto
sofiapinto.bsky.social
Sofia Pinto
@sofiapinto.bsky.social
Data science for social good @nestauk.bsky.social
Thank you for letting me know Jeremy!
November 17, 2024 at 9:32 AM
The goal of this publication was to share our method and get feedback on it, before we improve the existing profiles. Hence, the profiles we're sharing are not the final profiles.

🙏 Leave suggestions here: shorturl.at/Ojb4s
November 15, 2024 at 2:49 PM
👩‍💻 Data science work so far included:
- Defining smart meter data features that capture the main drivers of electricity usage
- Clustering households based on smart meter data features to identify electricity-use profiles
- Contextualising profiles with information about homes and households
November 15, 2024 at 2:47 PM
⏭️ The second phase of the work will focus on improving the existing profiles in response to stakeholder feedback and potential use-cases we identify. The final set of profiles will include gas consumption, alongside electricity.
November 15, 2024 at 2:42 PM
🏡 The goal was to understand whether smart meter data could be used to identify different patterns of energy usage and characteristics of homes and households associated with each usage pattern.
November 15, 2024 at 2:42 PM
🔋 In the first phase of our project, we have developed two prototypes based on half-hourly electricity smart meter data to group households based on how they use energy.
November 15, 2024 at 2:41 PM
We've analysed over 1 million forum interactions relating to home heating posted in the past 20 years.

🍃 Read our learnings here: lnkd.in/deX_SWNr
🖥️ Access our GitHub repository: lnkd.in/dtHXjG8p
🧑‍💻 Check our technical appendix: lnkd.in/dt_hkjum
LinkedIn
This link will take you to a page that’s not on LinkedIn
lnkd.in
October 28, 2024 at 4:49 PM
Thank you Marcus! Credit to @helloaidank.bsky.social as well, who delivered the data science with me.
October 25, 2024 at 3:57 PM
Findings can be explored in the dashboard at the end of the piece (you can see differences across profiles for different contextual features). Feel free to suggest additional contextual features you’d like to see in a next phase of the project, through our feedback form.
October 24, 2024 at 6:20 PM
Hello Mary - yes, we’re using a pseudonymised UPRN to join information about households in specific energy-use profiles with their respective EPC information. You can read more about how we do it in the technical appendix (it’s linked at the end of the project update page).
October 24, 2024 at 6:08 PM
Many people at Nesta provided support during this project.
Cath Sleeman, @lizgzil.bsky.social, @karlisk.bsky.social, Jack Vines, Daniel Lewis, @acjsissons.bsky.social, Adrian Stymne, Codrina Cretu, Andy Marsden, Oliver Zanetti,
Genna Barnett , Zayn Meghji, Elin Price and Robert Harris.
October 24, 2024 at 4:20 PM
🙌Acknowledgements:🙌
@helloaidank.bsky.social
for delivering the data science work with me!
October 24, 2024 at 4:19 PM
🔍Main takeaways:
- There has been an increased interest in heat pumps over time!
- Homeowners are still facing issues that are preventing them from getting a heat pump.
@nestauk.bsky.social's sustainable future team has been working across most of them!
October 24, 2024 at 4:16 PM
You can also see more information for some of the above topics. Cost can be split into several sub-themes including the upfront cost of installing a heat pump, the running costs of having a heat pump as a heating system and the related rise of electricity prices.
October 24, 2024 at 4:16 PM
Here we have the same for the second forum, which is more focused on self-building.
October 24, 2024 at 4:15 PM
The topics of conversation identified in the context of heat pump conversations and respective breakdown of sentiment, for one of the forums analysed - where we find mostly opinions from homeowners.
NB: those who are unhappy are more likely to share their negative experiences.
October 24, 2024 at 4:15 PM
🆕 Some of the learnings: 🆕
For most of the 2000s and 2010s, heat pumps were a niche topic on the forums we analysed. But interest has surged, and they were the 7th most discussed energy / heat topic in 2023.
October 24, 2024 at 4:15 PM
🤓The methodologies we used and the code we shared publicly can be applied to other domain areas/ other sources of forum/text data.
October 24, 2024 at 4:14 PM
To know more, you can:
- read our technical appendix: shorturl.at/A8bvk
- check our code through this GitHub repository: t.co/up7yjKYgRk
October 24, 2024 at 4:14 PM
📈 About the data science: 📈
We have identified topics of conversation using a topic modelling approach called BERTopic. We used sentiment analysis to identify the share of interactions that were positive, neutral or negative.
October 24, 2024 at 4:12 PM