Get Out - Team Kofte

• 

  Enter the date you are flying and where you are flying from.

• 

  We give you predictions on the amount of time it takes for security and gives you recommendations on which method of travel to use.

• 

  Don't worry, we have you covered with a survival guide!

• 

  Oh no, a nuclear attack!

• 

  Recommendations on other cities near you

• 

  Useful tools that you can order from amazon (We know they will still be making people work)

Inspiration

We were inspired to create this project because we believe travelling is an amazing experience, but sometimes organising a trip can be a hassle so we would like to help simplify and improve everyone's experiences.

Try it out yourself!

head over to https://guhgetout.netlify.app/

What it does

The website allows a customer who wants to rent a car to find out how long it will take to pass through security in the airport they are arriving at (with a mean squared error of 2.38) and how long it will take to get from the airport to the rental depot depending on the means of transportation they are going to use (with a mean squared error of 1.15). However, all of a sudden the Nuclear Fallout begins so an alert pops up on the screen and the user is being prompted with useful information on tips to survive: places to escape to and objects he could use for protection.

How we built it

We implemented a deep learning model for predicting the times in python using TensorFlow. Then we hosted the model on Google App Engine using a Flask server. The user interface was implemented in React and hosted on Netlify. Several API calls were made to convert the airport code into coordinates and city names. Additionally, we used external APIs to find closeby cities that we can recommend to the user.

Challenges we ran into

We had some challenges fixing CORS policy issues while trying to make the models into APIs and finding the optimal hyperparameters for the models. It was challenging to integrate a model from TensorFlow into a React app.

Accomplishments that we're proud of

We are proud of creating and hosting a web app, implementing regression models for predicting times, deploying the models and designing a creative user flow.

What we learned

We learned how to implement a regression model using deep learning and create an API using Google App Engine.

What's next for Get Out

• Training the model with more data for more accurate predictions
• Convert the Nuclear Fallout part of the website into a serious recommendations page for improving the customer's holiday
• Finding and using a better API for travel recommendations around the rental depot

Built With

• flask
• google-app-engine
• googlecolab
• jupyternotebook
• keras
• netlify
• python
• react
• tensorflow