Tech for Life on Land

Climate change has increased the frequency of natural disasters, yet the flow of information remains slow. Ekolog is a mobile-first solution designed to bridge the gap between an incident occurring and the data being analyzed.

Built in alignment with SDG 15 (Life on Land), the app empowers users to report environmental damage while using AI to instantly classify the severity of the event.

Subjective Reporting

When a user reports a flood or landslide, their description is often subjective. One person might call a small flood Severe, while another calls it Minor.

The Goal - We needed an objective way to categorize reports based on data (affected infrastructure, depth, area), not just feelings.

AI-Powered Classification

We didn't just build a form, we built a smart analyzer.

The Brain

I implemented a K-Means Clustering algorithm using Python. Unlike standard classification which needs labeled data, this unsupervised model analyzes historical disaster data to find natural groupings.

• It looks at factors like infrastructure damage and casualty count.

• It automatically clusters the report into High Severity or Low Severity buckets without human intervention.

Native Android Architecture

On the frontend, We used Kotlin and XML to build a robust native Android experience. The priority was speed—users in disaster zones need an interface that loads instantly and works reliably.

Serverless Backend

To handle real-time data without managing physical servers, We integrated Firebase.

• Authentication - Secure sign-up and login flows.

• Database - Storing reports and syncing them across devices.

Key Features

• Geo-Tagged Reporting - Users can submit photos and location data, which helps authorities map the spread of damage.

• Severity Auto-Tagging - The moment a report is submitted, the system tags it as High/Low Risk based on the K-Means model.

• Educational Hub - The app includes a section for Environmental Tips, educating users on how to prevent degradation before it happens.

Challenges

The biggest technical hurdle was bridging the gap between a Python-based Model and a Kotlin-based App.

• Integration Strategy - I learned how to decouple the Data Science model (which runs on Python) from the Mobile Client, using APIs to pass data back and forth.

• Real-World Data - Working with actual Indonesian disaster datasets taught me that real-world data is messy and requires significant preprocessing before it can be useful for AI.

Ekolog proved that mobile technology can be more than just social media, it can be a tool for environmental resilience. By automating the classification process, we created a proof-of-concept for faster, data-driven disaster response.