The travel problem your project solves, including TravelScrum challenge(s)
Difficulties in finding and receiving the latest safety updates during travel especially when pandemics, accidents or disasters happen. The safety information is scattered everywhere and takes time to reach us.
There are limited platforms that provide customized travel recommendations for every location and event.
There are too many manual steps to find out business-safe areas that makes business travel experiences complicated.
Your solution and what it does
Wanderer provides COVID-19 safety ranking in the scale of 1 to 5 (1 is least safest [Red Zone] & (5 is most safest [Green Zone]) of a location and recommendations of similar locations by studying other visitors' covid safety reviews, location, weather, time of the day and season using machine-learning technology. This information could be very important from visitor's standpoint as some visitors are more risk-averse than the others and they could apply these filters and only see the desired results.
A one-stop platform that will provide customized safety reviews of every event unique to each user. Safety ratings come from events such as viral diseases, tornadoes, flood, accidents, hazards, riots, wars etc. The important point here is that since we own the back end API we could fully change it to add more context related attributes that the customer wants for example age, gender etc.
Artificial Intelligence will digitalize the manual steps of corporate travel especially in TMCs that handle staff’s safety & risk factors. Wanderer will offer personalized business trips and preferences be studied to understand staff better. For example, in the future we could provide tags for Business safety ranking in the scale of 1 to 5, where business travelers can tag hotels that are located near to the police stations, have safe-lockers inside rooms, less pick-pocket crime reported areas, have proper emergency evacuation plans etc.
How did you build it?
This app is built entirely in python using some excellent python libraries - pandas, ski-kit-learn and Flask. Flask is used as a web front end to accept REST api based JSON messages while also sending JSON responses back. We currently only have a mock up for our UI as the UI engineer left our team for a different project, a day before the submission. But team did a great job mocking up the UI using Proto.io. Our dataset came from Kaggle and our backed was inspired by this great thread on the kaggle - https://www.kaggle.com/amiralisa/context-aware-recommender/comments.
What are you proud of?
The Machine-learning coding worked at the back-end and that we are able to scrum under the pressure to add the COVID-19 ratings as planned originally. Additional success of personalized users' data which would be useful even after the pandemic period as well as beneficial for the corporate travel industry.
Team members cooperated well even though we had to let go our one and only front-end developer. Everyone gave their best efforts and positive spirit continues.
What is next step for your solution and how will you take that step?
Application - We planned to enhance to a full working app by completing the front-end development. We will also add options to add reviews using texts, photos and videos along with the safety star ratings. Mainly we would like make the app very simple to use, neat and reduce number of frictions like asking too many inputs from the user to study their preferences instead we just study from user's activity like places someone felt more threatening than for others, timings of the day we get low safety rating from users (like for a woman night time would be less safe in some regions), study crime reviews reported and be more useful to the corporate travel industry by solving more safety and risk related issues.
Business - We planned to do marketing and get more users sign up on the basis of B2B2C business model where we can focus on complementing travel management and corporate travel companies. Wanderer will be a complementing engine to Google, TripAdvisor, Waze and travel platforms of airlines, TAs, OTAs and DMOs in the future – We can save money on customer acquisition cost! Our API just adds machine learning intelligence to your already available APIs. We know that every successful travel websites have tens of APIs (if not hundreds) and we don't want to re-engineer or duplicate the work. You can still have all the safety ratings that comes from your other APIs (and our special Covid-19 Rating) and other hazards, riots API. We apply intelligence to it.
What tech did you use to build it? (e.g., APIs, tools, other relevant details)
flask, machine-learning, pandas, python
Links to test the solution (e.g., Github, Website, App, adobe)
Working Code Demo: https://github.com/vinodhalaharvi/Wanderer
UX Demo: https://pr.to/BADLK5/
Outro - Light on user-personalization
Have your ever visited Amazon or Netflix (just kidding) and have you ever visited Walmart (also kidding) or other sites lately? Have you noticed any difference in experience and wondered why sites like Amazon keep you coming for more, want you buy stuff that you don't really need but you wanted it and now you have it :)? Have you ever ran out of movies to watch and asked a friend for his faves list so you could just steal watch them? Wonder why? why? well the answer is user personalization. When you enter amazon's site they know more about you than you think. They know more about users like you! they know what you don't know, what you are about to know and what you are about to make other people know what you are going to know. It's really that simple. Knowing your user and giving suggestions after you know your user make a lot of difference, or just enough to have them come back again. That's a business winning decision. From the user's perspective they want to know what has been "collective" common sense. We think this is how all the travel sites should be, so if you need to find a fun place for your next visit you don't have to google it. It should be obvious - at least after we tell you :).