At the recent Think Data for Government event, the Met Office and Made Tech shared the weather service’s journey from forms and spreadsheets a fully automated, machine learning-enhanced data processing pipeline.
By automating the sentiment analysis of vast amounts of user feedback, the Met Office could extract patterns and trends with an efficiency that was previously impossible. Combined with traditional methods of user research, the company was able to provide in-depth, qualitative insights into user behaviours.
“[We were] able to start looking at whether the feedback we’re getting was positive or negative, and be able to put those topics together and start seeing where we’ve got issues, where we don’t have issues, and where we’ve got areas that we could develop,” said Mat Gard, solutions architect, Met Office.
The result has been a faster response to user feedback, allowing for quicker testing and implementation of new features.
“Hopefully [users] see the results of their feedback coming into the app in a much more fleeted foot manner. We’re able to start testing new features. We’ve got a pipeline where we can push stuff out to them for testing new maps, for instance, we got those coming later on this year. Things like that, where we can actually build a relationship with our user base,” said Gard.
“The speed of response is really noticeable, closing that feedback loop from a user raising an issue or a bug in the comments. That [is] raised with the development team…a few weeks, months before the regular alarm bells will start to ring. So that’s been really great to see,” said Made Tech’s lead data scientist, James Poulten.
Poulten also highlighted the Met Office’s ability now to “filter out some of the noise.” He said it could better “observe trends in data, the aggregated statistics, rather than focusing on small comments or small sections of comments when you have three, four, five users talking about one particular issue that could set the start alarm bells ringing until you realise that that’s five comments in a sea of 10,000 comments that day.”
If you liked this content…
“It really helps to highlight the important things, rather than the noise of things,” agreed Gard.
Building relationships with the data
Gard insisted that the partnership between the Met Office and Made Tech is key.
“The Met Office has a great amount of scientific knowledge, it has a great amount of technological global knowledge, it has a great amount of product development knowledge. But it didn’t have a great amount of knowledge around how to pull that data together from a customer perspective and be able to offer something new. And for me, that was where this really helped drive that home.”
He continued: “Obviously with this data we can build relationships internally within the Met Office with relevant science teams and other technology teams, and also with our steering customer groups, so that we can go to them and say, ‘these are the things that our public actually want, and these are the things that we’re doing to deliver that for them’ So we can actually show value for the money that we are being given.”
“Working with colleagues at Met Office, everyone there is very passionate about what they do. They’re really deeply invested in producing accurate weather data for general public, and there’s so much knowledge that we’re now able to back up with user feedback,” added Poulten.
“For example, off this work we’re starting a piece around the perception of precipitation – the idea that just because we’re saying it’s going to be showering, doesn’t mean people necessarily understand why it might not shower. So…people don’t really understand how we have the data to back up that user story, and we can build a business case around doing some work to remedy that and fix it.”
