Covid Trip Planner - COVID-19 risk forecasts for 700+ destinations worldwide

@kristinides Thank you Kristin! :)

@kristinides That's a nice description. Hope we don't need to use it so much. Thanks!

@prilutskiy Thanks!

I travel very often and I have no problems with the bank. Seriously, it seems to me that you are being deceived and you should look at other banks. I found mine when my friend recommended me a list with cool travel apps. Travel apps are a very cool thing! The opportunity to travel is always great. Especially now, when smart apps simplify our lives and help us buy tickets, book hotels, easily navigate unfamiliar places, plan routes step by step and choose the most gastronomic places. Previously, this required thick travel guides, but now only your smartphone is needed. It helped me in this dbfahrplanauskunft.com since I can see all the planned trips there, find out the prices and buy tickets in advance! I really like it! I am very glad that I can now plan my trips on my own.

@nazli_ozkan1 Thanks a lot for your support! I agree, although it is not recommended, people still have to travel, and while making decisions this tool will definitely be useful.

@nazli_ozkan1 Great question, thank you! You can think of our Machine Learning model as an oracle that predicts next week's weekly case count for a destination, given its weekly case counts for the past few weeks. We then "feed" next week's prediction back into the model and predict how the situation will look like in 2 week's time. Currently we go up to 8 iterations like this to predict the next 8 weeks. This model was trained with past data from all destinations across the world. We then use these predictions to estimate how UK/US based restrictions may change in the future, based on each country's own definition.

@arda_de Thanks for your interest! We are using stacked LSTM networks as the predictive model. Accuracy is hard to quantify at this stage, as all we have right now is the model's error/loss value across the samples. However, we're planning to put a dashboard to keep track of the accuracy over the weeks, which should help in determining whether the models are getting better over time or not. We make 100 predictions per destination using slightly different parameters to learn their spread (uncertainty) at any point in time. If you take a look at any destination's information card, you will see that there is a shaded area around the predictions. Higher spread means lower confidence. The covid-case data is fetched from here (JHU): https://github.com/CSSEGISandDat.... The travel restrictions are from UK/US official travel advice websites. In cases where the case counts do not match the official travel advice, we put a note in that country's information card that says "[CTP: Non-Covid factors may be at play. Check link for details.]". The advice link contains why this country is treated differently. Travel advice depends on other factors such as security and politics as well. Hope this info was useful! We're planning on releasing the code soon, stay tuned :)