EcEcursion

Taipei | You Are My Sunshine

POWER Data Interface and Prediction System (POWER_Dips)

High-Level Project Summary

We created a cross-platform app for the NASA POWER API providing a better and more intuitive mobile interface while predicting future trends of Photosynthetically Active Radiation using Machine Learning techniques, which may help estimate future solar power harvesting. As household solar power harvesting becomes more and more popular we consider our ML application of high importance and high demand. Pure data-based applications tend to be less attractive and less used, thus with refined visuals and an intuitive interface we aim to create a better user experience for both transitioning and new users while being able to provide processed and ready-to-use information for better accessibility.

Link to Project "Demo"

https://drive.google.com/file/d/1hm_TSv-PzHzCPH-fycysvtegK9e2x0tc/view?usp=sharing

Link to Final Project

https://github.com/NASA2021-EcEcursion/

Detailed Project Description

*.apk files for immediate download and run could be found at Releases tab in given GitHub*

In this task we used React Native to create a cross-platform application, replicating the POWER API web-based interface for easier use on mobile devices while also adding on some of our own, making use of the data and creating the app not only as a data retrieving and visualization tool, but able to be used more often having better functionality. The user flow is mostly the same, however we also made some tweaks regarding the interface to give both transitioning and new users a better experience.

Google Places API is used for locating, enabling non-Longitude/Latitude based positioning for more intuitive use by the mass public. Also, the data retrieved could be viewed as better-looking graphs to ensure accessibility. Furthermore, we wrote a Python backend hosted on Azure for LSTM prediction on values of Photosynthetically Active Radiation, which helps estimation of future solar power harvesting for households.

This is a relatively complex project as we incorporated many APIs and tools: Microsoft Azure ML Studio and Python Web App Hosting, Google Places API, Facebook React Native, Flask, RESTful API, Keras, TensorFlow/TensorFlow Lite , GitHub and POWER API by NASA. The main coding languages are Python and JavaScript.

We aim to make data more approachable and enjoyable by making the retrieving process more intuitive, creating better visuals and incorporating real-life applications. As household solar power harvesting becomes more and more popular, we consider our ML application having high importance and of high demand.

Space Agency Data

NASA Power API is used in our work. ALLSKY_SFC_PAR_TOT is used in ML training in LSTM.

Other data from numerous space agencies regarding Sunlight data is referenced while research on ML inputs for correlation determination.

Hackathon Journey

This is the first Space Apps experience for our team, and it was definitely quite a pleasant experience. We all came from different backgrounds and have different expertise, and working with a group having such a diverse skillset is not an easy task as we often don't share the same opinion. However, through the process we think we all learned a lot on teamwork and communication. "You Are My Sunshine" is our challenge of choice because we consider that this task could bring out the best of us utilizing almost all our skillsets. Some of us were forced to learn new frameworks while all of us contributed to this work. Consisting of a group of engineering majors, we may not have the flashiest visuals but set our eyes on and surely tried hard to bring a pleasant experience to the users using our app while also bringing additional features to the table.

The whole process was challenging as it was extremely difficult to create a fully functional app within 48 hours with our skillsets being so scattered. We had a lot of trouble connecting all the pieces together as a lot of different frameworks and interfaces were used, and we were forced to play around with things until everything ticks and all finally comes together.

Here we have to thank the organizers for all the help and all the hard work conducting such an event, especially due to the pandemic situation and being in an all-virtual format makes the logistics even harder. We felt comfortable during the whole process despite being in lack of sleep these few days, thanks a lot :)