Machine Learning Model for Flood Detection in Karachi banner

Machine Learning Model for Flood Detection in Karachi

Created by: eshangillani1
7 devlogs
14h 7m 33s

Updated Project: Fixed the demo by adding a front-end using Node.js

My main focus for this project was building the machine learning model so the front-end is not perfect due to it being the first time I'm attempting to build/deploy a front-end…

Updated Project: Fixed the demo by adding a front-end using Node.js

My main focus for this project was building the machine learning model so the front-end is not perfect due to it being the first time I’m attempting to build/deploy a front-end, but also because my project was rejected multiple times for my demo of normal Python script not being able to run.

This project was inspired by the flooding in Pakistan in 2025, and it’s an early warning system. It uses the Random Forest machine learning model in order to make these predictions. It was trained on humidity, atmospheric pressure, and precipitation for the last three months as well as actual flood weather records in Karachi. It makes live predictions by taking live, accurate weather data from the OpenMeteo API.

This project uses AI

AI used for basic API calling debugs, and to help develop the front-end and deploy it, using Claude Code in order to have a visual demo rather than the previous Python run.

Demo Repository

README

Loading README...
eshangillani1
eshangillani1 worked on Machine Learning Model for Flood Detection in Karachi
14 days ago
3h 18m logged

Worked on further software improvements on the frontend as well as working towards implementation of more cities and locations on the backend, although it isn’t visible on the frontend just yet due to the web code needing to be updated!

Attachment
0
eshangillani1
eshangillani1 shipped Machine Learning Model for Flood Detection in Karachi
28 days ago

Shipped this project!

Fixed some more issues with style and graphs on the front-end, specifically on the map of the region. In the future, I would like to make this a Google Maps and connect this machine learning model to cities across the world so that more people feel safe and secure knowing that they’ll be prepared for emergency flood events.

eshangillani1
eshangillani1 worked on Machine Learning Model for Flood Detection in Karachi
28 days ago
0h 47m logged

Worked on improving clarity, graph accuracy and style consistency throughout the front-end. I also worked on further training the random forest model with data from further back in time (around 6 months rather than the previous 3 using the OpenMeteo API.

Attachment
0
eshangillani1
eshangillani1 shipped Machine Learning Model for Flood Detection in Karachi
29 days ago

Shipped this project!

Hours: 9.34
Cookies: 🍪 36
Multiplier: 3.83 cookies/hr

I developed a front and back-end for my Machine Learning model using a Random Forest in order to predict the chances of flooding occuring in Karachi, Pakistan. I finally have a solid UI front-end as well as an accurate backend as well.

eshangillani1
eshangillani1 worked on Machine Learning Model for Flood Detection in Karachi
about 1 month ago
4h 26m logged

Worked on developing a front-end for my Machine Learning model. I used Vercel for deployment and used Node.js for the frontend. Currently, it just shows the latest status, but I plan to add more in the future such as weather statistics and graphs

Attachment
1

Comments

eshangillani1
eshangillani1 about 1 month ago

The issue regarding the failed deployment has been corrected.

eshangillani1
eshangillani1 worked on Machine Learning Model for Flood Detection in Karachi
about 2 months ago
3h 16m logged

Worked on adjustments to the structure to make it more accessible and easy to replicate for different cities, as well as being able to expand the number of natural disasters covered. Additionally, worked on getting it to PyPI for the demo

Attachment
0
eshangillani1
eshangillani1 worked on Machine Learning Model for Flood Detection in Karachi
2 months ago
0h 46m logged

Was working on moving to PyCharm and making basic adjustments for functionality and ease of use. The API pull is still quite slow when running it for the first time, taking around 2-3 minutes, but that seems to be something out of control.

0