Thanks, I am working on that, would be very happy if you could mention the biggest obstacle you faced. Also initial use case is very specific, which affected a lot of first decisions.
Thanks for the provided details and a report, I did try it in couple of places outside UK and outside of 4G coverage, but I will try more options via VPN soon. Did you use specific device or browser? There was small outage this morning which could have affected when you tried it this morning.
You got it right, thanks for taking time to explore it. I definitely need to add some sort of wizard, if you remember a very good one from other application you recently use - I would be happy to look at it for an inspiration.
Hello, this is an application I am working on and here's my story.
It is a place exploration tool, which allows to explore globe by favourite weather. For example if you liked Barcelona in June last time, it could show where else would weather be similar to that in January, when you have your travelling planned.
Back in December during Christmas break I wanted to check where else is as nice as during my last holidays. I did not find a tool to do that and I wanted to learn some functional stream processing as well (I am a developer), so I took it as a weekend challenge to see if it's possible to quickly model at all. A weekend project grew into a week, then a month, one thing after the other and now here I am with Flycatcher Club.
Currently I am considering only temperature (but have other weather data too and working on integrating it soon - especially precipitation patterns). Regardless of this lacking it is already very accurate as is.
My biggest concern was that because of differences each year the historical data will be complete garbage for matching, which proved not to be true. I periodically run benchmark by processing all the data up to a year back and comparing matches to past year's measurements. There is small number of extreme outliers of course, but 70% of matches are within 3 degrees Celsius (which I consider accurate enough) with the average accuracy being within 4 degrees across all data.
I would like to integrate other data too, so that people could start filtering weather matches by cultural, economical, reachability and other criteria. If you know some available datasets or have in mind what you would find useful - please let me know.