AI network to predict the next drought and flood of an area.

Inspiration

In todays world, the dangers of natural disasters has been downplayed due to our current circumstances. The truth is, natural disasters have slowly become a massive problem. Statistics show that the number of natural disasters has been increasing since the 1960s and a sharp 35% from 1990. The combination of overpopulation, pollution and global warming is the cause. 60,000 people every year die from natural disasters which we believe could have been prevented.

What it does

Our HTML website was built to make the public more aware of common natural disasters and how to prepare and survive them. The main part is the python code. The python code uses data and analyzes the data to make a prediction for when the next drought or flood would be.

How I built it

The HTML website was made with the ideals of simplicity and accessibility in mind. We wanted it to be easy to use and navigate so the user gets the most of this website. CSS was integrated into the website in order to make it look cleaner and more appealing. Flask was used to integrate the python code into the HTML website as well as the HTML website to navigate itself. The python code was made to analyze data effectively and efficiently. To predict we used a Line regression algorithm in Machine learning to get our prediction. Factors such as precipitation, humidity, air temperatures, El Nino, ocean temperature, and more were used in the algorithm. The algorithm itself was made with a linear regression inspiration. This was further advanced by using statistical techniques such as a line of best fit to have greater accuracy. We used bots to optimize the code. The data itself was gathered by web scraping. This is the process by which we use the requests library to directly interact with a website. It was helpful in making the code much less cluttered and using a lot more data.

Challenges I ran into

A challenge we ran into was successfully finding a way to have a python backend with a HTML front end but nothing seemed to work. We tried flask, Django, and some others but to no avail. We struggled with this for a while. We finally as a group figured out that we could just make a new python project and have our HTML files in there. This is where we applied flask and thankfully it all worked out.

Accomplishments that I'm proud of

Something we are proud of is the accuracy of our algorithm. We did not expect the predictions to be too exact maybe a 3-month margin of error but there was barely any. We used old data in order to "predict" a drought that we know for a fact happened and the prediction was spot on. We tried this for our flood prediction algorithm and thought it was as accurate, it was still quite close.

What I learned

We learned a lot of new concepts. Some examples are machine learning, a concept we had not dabbled in too much as well as flask. It was interesting how we could in a way mimic a human mind to solve our problems. This project not only taught us new topics but also tied a lot of information together. We have all had experience with python and HTML but separately. We did not know there was a way we could have them work together in a way to create a product. It truly was in a way a beautiful experience.

What's next for AI network to predict the next drought and flood of an area.

As our plan went quite successfully, we plan to pursue this project in the future even after this hackathon. We want to include a lot more states than Louisiana and California first and foremost. When we are able to handle that data well. We want to try to take into account more factors that could influence the prediction. We also want to look at other kinds of natural disasters that are harder to predict (such as earthquakes) and find a solution.