predictor

Inspiration

I am inspired by Analytics Vidhya, who did a case study on predicting wildfires. Furthermore, having lived in Australia for 5 years, which is an region that has heavy rains and wildfires, I believe that it is essential that the process of responding to such weather events could be sped up and made more effective. This could also be used in synergy with local methods, such as the Aboriginal method of ‘cool burning’, which can both control wildfires and help the land grow.

What it does

It takes in a dataset and outputs several things:

The correlation graph. This shows the relationship between numerical variables, which can be used to source factors that affect a certain weather event.
The results. This is what the machine predicts for each entry, and this can be used to manage and prepare for events.
The confusion matrix. This shows the algorithm’s correct and missed attempts. This could be used to improve the algorithm.

How we built it

Predictor is built on streamlit, which is a platform built for data science apps. It is built using python, and using data science libraries such as pandas and scikit-learn.

Challenges we ran into

Originally, the fucntion that the model uses failed to run. After several hours, it is finally fixed. Furthermore, streamlit's file_uploader command does not return the csv file in a way that could be read by the pandas.read_csv command.

Accomplishments that we're proud of

I am proud that the project is functional, and that it has a high accuracy of 84.69% (to 2 d.p.).

What we learned

I learned to first plan out the project, exploring all of its possibilities. Most importantly, (no matter how cliché it might sound) to not give up too quickly, as a few hours ago this was going to be another scrapped project.