Really interesting approach! I can see this being useful. How are you dealing with short/medium-term changes in consumer sentiment, I assume your model is currently fairly static? For example, the results to "Would you buy an e-bike?" might change over time as cities add charge-points or additional bike paths, prices for e-bikes go down, etc. And as a more extreme example, the answer to typical YouGov questions like "Who will you vote for in the next presidential election" will obviously change daily based on a multitude of factors that aren't present in your training data.
The data we trained on has year, so we can specify the year you ask the question (the default is 2023). You can also see how answers change over time. [1] shows how the distribution for "Do you support the President" changes from 2000 to 2023 (see the 9/11 spike, end of Bush era, Obama era, Trump era, etc.)
