search

Capstone-Projects-2021-Fall / project-proposal-4398-sections04-flagg

project-proposal-4398-sections04-flagg created by GitHub Classroom
1 stars 0 forks source link

Better Weather #6

Open gannontraynor opened 3 years ago

gannontraynor commented 3 years ago

Project Abstract

Modern weather reports are made every morning by sending up a weather balloon into the atmosphere to make a bunch of recordings and then meteorologists who have studied their craft for years make an educated guess that the rest of the world then plans their day around. This application will instead make a weather prediction based on historical data and a machine learning algorithm cutting out the possibility of human error.

High-Level Requirement

Mobile/Web application that utilizes machine learning to predict the weather better than humans. (Hopefully)

Conceptual Design

This is a mobile application (Kotlin and Android Studio for Android) (Xcode and Swift for apple) that is going to execute a machine learning algorithm, (built-in python) that will take a set of parameters like wind, barometric pressure, temperature and produce a weather forecast.

Background

Traditionally, the weather is one of the few things in this world that can't be predicted, however, the advancing technologies in machine learning are beginning to challenge that claim.

Required Resources

Solid understanding of machine learning principles in python and the JupyterLab notebook. A computer with a dedicated RTX GPU is preferred but not required.

Supplemenal

https://towardsdatascience.com/weather-forecasting-with-machine-learning-using-python-55e90c346647

Xhuljano-Racaku commented 3 years ago

This is on top 3 project that i choose. It looks interesting, and I love anything related to weather, so this will be a plus. I have used both jupyterlab notebook and python. I am thinking we can do something for the users are be able to monitor specific events that directly affect them while also seeing what is affecting the rest of the world. I haven't done a lot of machine learning, but I am willing to learn more during this project

tuj39467 commented 3 years ago

I like this idea a lot because weather uncertainty is one of those inconveniences that can either make or break someone’s day. I can’t count the number of times where I’ve walked around campus without bringing an umbrella thinking there’s no chance of raining and then it starts pouring really hard unexpectedly.

I have programmed in the past using both python and Android Studio so I can contribute in those ways. However, I have never used JupyterLab Notebook but I am willing to put in the work to understand it well. I am also willing to contribute to this application by making it visually appealing and user friendly.

Phaceial commented 3 years ago

Everybody interacts with the weather. I think the project has a wide reach and can be really impactful. Love the idea, but here's a thought. Since it seems like you're going to have a server processing data, do you need the apps to be natively developed? When I think of the headache involved in developing for two separate platforms it's necessary when collecting data from with the devices or resource intensive tasks. If you're offloading that somewhere else to run a python script you can likely just use react native and save time on testing two completely different code bases.

I have a lot of experience with mobile development for Android. Not so much with Kotlin or iOS development. Always been interested in machine learning, but I have never played around with it. My only Python experience is writing scripts to do my math homework and writing a discord bot. Interested to see if this project is picked. Good luck!

Eriksiano commented 3 years ago

Although, I do not know much about machine learning principles I would like to contribute to this project because It is something that will improve with time as more weather information is recorded. I think we can achieve fairly accurate results with data that is currently available.

I would like to contribute to creating an appealing user interface, especially since I have some experience with Android Studio. Although, I would recommend making this project as a web based application with Python/Django or some other Python framework to keep the technologies used in the same environment so contributors have an easier time learning similar technologies that are still meaningful.

gannontraynor commented 3 years ago

I like this idea a lot because weather uncertainty is one of those inconveniences that can either make or break someone’s day. I can’t count the number of times where I’ve walked around campus without bringing an umbrella thinking there’s no chance of raining and then it starts pouring really hard unexpectedly.

I have programmed in the past using both python and Android Studio so I can contribute in those ways. However, I have never used JupyterLab Notebook but I am willing to put in the work to understand it well. I am also willing to contribute to this application by making it visually appealing and user friendly.

I will need all of the help in the world with the UX/UI design, Im a lost soul in that world.

gannontraynor commented 3 years ago

Everybody interacts with the weather. I think the project has a wide reach and can be really impactful. Love the idea, but here's a thought. Since it seems like you're going to have a server processing data, do you need the apps to be natively developed? When I think of the headache involved in developing for two separate platforms it's necessary when collecting data from with the devices or resource intensive tasks. If you're offloading that somewhere else to run a python script you can likely just use react native and save time on testing two completely different code bases.

I have a lot of experience with mobile development for Android. Not so much with Kotlin or iOS development. Always been interested in machine learning, but I have never played around with it. My only Python experience is writing scripts to do my math homework and writing a discord bot. Interested to see if this project is picked. Good luck!

Kotlin is a new language from Google specifically for Android development, it takes the best of python and java and combines them together, or so I'm this will be my first time using it. And yeah, the only thing the apps will have to do is access a server and the phone's location.

KarlSchaller commented 3 years ago

To me, this is a really cool potential application of machine learning. I'm always interested in new ways of applying AI, and weather prediction seems like a field just waiting to be taken over since a long history of data is public and easily accessible.

I have a good amount of experience with machine learning, so I should be able to contribute on that front. One potential issue that might arise is climate change. It's possible that the decades of historical data that we would want to use for training are no longer as applicable as we would like for making predictions for today because of the amount that the climate has changed in recent times. Or it's also possible that this won't be an issue at all if the relationships between the variables have stayed the same during this time (at the end of the day I have no idea how predicting weather works). I also have experience with mobile application development, so I could help out there too, but it sounds like the UI might not need to be that complicated.

AidanCodes commented 3 years ago

I think this is a great idea for an app and will be a great challenge to work on! It seems like a great opportunity to work with machine learning and utilize it to make someone's day better.

Python has never been my main language, but I do use it on a regular basis for work-related tasks (PyQGIS specifically). I could definitely contribute on the Python side and I am a quick learner for the JupityrLab stuff. Most of my mobile dev experience is with the older Java-based android flow, but I am happy to update my skills as needed.

One thing I wonder is if we will be required to pay to access weather data of if we can really get everything we need for free. I notice that nearly all mobile weather apps have either filled up with ads or require a subscription, and I am concerned that there may be a reason for that. I hope we can find some creative solutions for filling in the gaps and using live data as well, as I am always looking for new ways to keep my data up to date!

gannontraynor commented 3 years ago

To me, this is a really cool potential application of machine learning. I'm always interested in new ways of applying AI, and weather prediction seems like a field just waiting to be taken over since a long history of data is public and easily accessible.

I have a good amount of experience with machine learning, so I should be able to contribute on that front. One potential issue that might arise is climate change. It's possible that the decades of historical data that we would want to use for training are no longer as applicable as we would like for making predictions for today because of the amount that the climate has changed in recent times. Or it's also possible that this won't be an issue at all if the relationships between the variables have stayed the same during this time (at the end of the day I have no idea how predicting weather works). I also have experience with mobile application development, so I could help out there too, but it sounds like the UI might not need to be that complicated.

The climate change thought is super interesting. I don't anticipate that being a problem because climate change is a constant gradual change. We might be able to use the model to predict the severity of global warming in the next 5/10/100 years. That's definitely going to be very interesting to explore regardless.

Wyllice commented 3 years ago

Machine learning was my favorite computer science elective course, and I would very much enjoy working on a project that involves machine learning. The challenge of working on predicting weather is an especially interesting problem since we are all affected by the weather, and we all know how difficult it to make predictions.

I think my experience from the machine learning course and my general knowledge of Python can help contribute to the weather predicting algorithm and its implementation. I don’t have any mobile app development experience, but I’d be happy to work hard to learn what is necessary for the scope of this project. Would be especially interested in learning Swift as I have experience with Xcode and Apple products generally.

daviddcho commented 3 years ago

A quick story: During the summer, I planned on going to New York for a day with some friends but then the weather app said it was going to rain all day there. So we postponed our plans to another day. I wake up on the day we originally planned, and the weather app said it wasn't raining anymore. It would've been nice if the weather app was accurate earlier!

So I think it would be fun to learn about how weather is predicted and how it can be improved. I have experience with machine learning and Python to help contribute towards a hopefully more accurate weather prediction. I've also been looking into Kalman filters over the summer and I wonder that can be combined with machine learning to help with predictions. I don't have any mobile app development experience but I wouldn't mind learning about that to help create a nice app.