BantayAyuda: AI-Driven DSWD Emergency Cash Transfer System

Web App Repository: https://github.com/YashleyChua/BantayAyuda

Overall Champion at Hackamare: AI Hackathon 2025 (iACADEMY Makati), themed Innovating with LLMs. Developed by TryCatchers, Developers Society (DevSoc) Benilde.

---

Project Description

BantayAyuda is a Django-based AI web application that automates Emergency Cash Transfer (ECT) distribution for flood-affected communities. Combining machine learning, geospatial mapping, and SMS alert integration, it enhances the speed, transparency, and accuracy of post-disaster financial assistance. Designed for LGUs, DSWD personnel, and residents in urban flood-prone areas, the system reduces traditional needs assessment processes from weeks to hours.

---

Objectives

On November 13, 2025, during iACADEMY Makati's Hackamare: AI Hackathon 2025, our team developed an AI-powered workflow that streamlines disaster damage assessments and automates cash assistance evaluation. BantayAyuda supports disaster-response units by improving speed, reducing human error, and ensuring equitable assistance distribution.

---

Problem Statement

Flood-affected communities often suffer from:

• Slow, manual post-disaster needs assessments (PDNA)
• Inaccurate or inconsistent damage classification
• Delayed release of emergency cash transfers
• Lack of transparency in beneficiary selection and disbursement

Frequent typhoons in the Philippines cause widespread displacement, structural damage, and income loss. Improving data-driven relief allocation is critical for community resilience.

Target Users: Local Government Units (LGUs), Department of Social Welfare and Development (DSWD) administrators, and residents eligible for ECT.

---

Solution Pathways

BantayAyuda improves disaster-response workflows by:

• Classifying household damage levels using ML.
• Automatically determining cash assistance amounts (₱0 / ₱5,000 / ₱10,000).
• Mapping flood zones in real time through GIS.
• Sending simulated SMS qualification updates.
• Providing admin dashboards for LGUs/DSWD.

Solution Features:

Feature Type	Feature Description
Core	Damage classification (none, partial, total) via CatBoost; Automated ECT computation
Enabling	Prototype SMS alerts; GIS hazard mapping (Leaflet.js + OpenStreetMap)
Enhancing	Django Admin dashboard, real-time visualization, interpretable ML metrics

---

Machine Learning Model

• Built using CatBoost Gradient Boosted Decision Trees (GBDT).

• Achieved 100% accuracy, 100% precision, 100% recall, and 1.00 F1 score on structured theoretical data.

• Damage categories:

  • 0 — No Damage
  • 5000 — Partial Damage
  • 10000 — Total Damage

To refine classifications, the model includes a flood depth : house damage ratio, comparing:

• measured / estimated flood height (m or cm)
• house structural resistance (materials, elevation, wall type)
• predefined hazard thresholds based on PDNA guidelines

This ratio improves reliability in distinguishing partial vs. total damage.

Model Inputs Include:

• Flood depth & duration
• Type of housing material (GI sheets, concrete, light materials)
• Geolocation proximity to floodplain
• Flood:House damage ratio

Integrated directly into Django through .bin model loading for real-time inference.

---

Data Folder Generation

Note: The contents of the data/ folder (including waterlevel and household datasets) are automatically generated by running the ETL scripts:

• python etl/scrape_waterlevel.py
• python etl/scrape_households.py

Simply run the above scripts and the required data will be created in the appropriate subfolders.

---

Project Structure

ect_allocation_model/
├── etl/
│   ├── scrape_waterlevel.py      # ETL: DOST-ASTI flood sensor data
│   └── scrape_households.py      # ETL: Microsoft Building Footprints
├── notebooks/
│   └── ect_allocation_model.ipynb # Main ML training & evaluation notebook
├── models/
│   └── *.bin                      # Saved CatBoost model binaries
├── data/
│   ├── waterlevel/                # Cached flood sensor CSV data
│   └── households/                # Cached building footprint GeoJSONL data
├── .gitignore
└── README.md

---

ETL Pipeline

The model is powered by two automated ETL (Extract, Transform, Load) modules that scrape real Philippine government and open-source geospatial data.

etl/scrape_waterlevel.py — Flood Sensor Data

Detail	Description
Source	DOST-ASTI PhilSensors — real-time water level monitoring stations across the Philippines
Method	Scrapes the DOST-ASTI monitoring endpoint via session-authenticated POST request with CSRF token extraction
Output	df_hist (raw sensor readings per location/hour) and df_hist_agg (aggregated mean flood_depth_m and sensor_flood_depth_m per location)
Columns	Location, Province, Region, sensor_flood_depth_m, flood_depth_m, hourly readings, date
Coverage	~43 flood monitoring stations nationwide
Caching	Timestamped CSV files saved to data/waterlevel/
Fallback	Raises ValueError if scraping fails — no synthetic data generation

Key Functions:

• run_dost_etl() — Main ETL pipeline; scrapes sensor data and validates required columns
• load_and_aggregate_flood_data() — Loads historical data and aggregates by location
• refresh_realtime_waterlevel() — On-demand cache refresh with latest sensor readings

etl/scrape_households.py — Building Footprint Data

Detail	Description
Source	Microsoft Global ML Building Footprints — philippines.geojsonl.zip (~5 GB, 17.4M building polygons)
Method	Downloads the GeoJSONL archive, extracts it, and streams features using Fiona with spatial bbox filtering for efficient memory usage
Output	household_df with Latitude, Longitude, House_Surface_Area_Sqm per building
Bbox Filter	Metro Manila region: (120.85, 14.45, 121.15, 14.85) — yields ~922,905 building footprints
Area Calculation	Geometries projected to EPSG:3857 for accurate area computation in square meters
Caching	Raw GeoJSONL cached to data/households/; re-extracted from zip on first run
Fallback	Manual line-by-line GeoJSON parser if Fiona/GeoPandas unavailable

Key Functions:

• load_household_measurements() — Main entry point; supports bbox, cache_dir, and max_rows parameters
• _load_geojsonl_with_geopandas() — Fiona-based streaming loader with spatial filtering
• _load_geojsonl_manual() — Fallback line-by-line parser with centroid/area calculation
• generate_base_training_data() — Combines flood locations with building footprints, assigns Damage_Classification based on flood-depth-to-height ratio

Training Data Generation

generate_base_training_data() merges the two ETL outputs:

1. Tiles flood sensor locations cyclically to match the number of building footprints
2. Derives House_Height from sqrt(House_Surface_Area_Sqm) as a structural proxy
3. Computes Damage_Classification (Not Damaged / Partially Damaged / Totally Damaged) using flood-depth-to-height ratio thresholds (0.4 and 0.8)
4. Produces a training DataFrame with columns: Location, Latitude, Longitude, Flood_Depth_Meters, House_Surface_Area_Sqm, Sensor_Flood_Depth_Meters, Damage_Classification

Feature engineering (Spark SQL) then adds Flood_Height_Ratio, Flood_House_Damage_Ratio, and the target variable Target_ECT_Amount.

---

Tech Stack

Category	Technology / Library
Backend	Python, Django, Django REST Framework
Frontend	HTML, CSS, JavaScript, Leaflet.js
Machine Learning	CatBoost, scikit-learn, pandas, PySpark
Geospatial	GeoPandas, Fiona, Shapely
Data Sources	DOST-ASTI PhilSensors, Microsoft Building Footprints
Database	SQLite3
GIS Mapping	OpenStreetMap, Leaflet.js
Automation	Prototype SMS Integration
Dashboard	Django Admin
Visualization	Seaborn, Matplotlib

---

Key Takeaways

• Automated PDNA reduces response time and improves consistency.
• GIS layers help LGUs visualize hazard zones immediately.
• ML-driven classification ensures fair and transparent ECT distribution.
• Scalable architecture can be deployed nationwide for future typhoon seasons.

---

System Demo

---

Team TryCatchers – Roles and Responsibilities

Name	Role
Gabrielle Ysabel Almirol	Machine Learning Developer, Front-End & Back-End Developer
Yashley Joaquin Chua	Researcher, Front-End & Back-End Developer
Xander Julius Abo	Researcher, Front-End Developer

Technical Contributions

• Built an automated ETL pipeline web-scraping real-time flood sensor data from DOST-ASTI PhilSensors (43 stations) and streaming 922,905 Metro Manila building footprints from Microsoft Global ML Building Footprints (17.4M features, ~5 GB GeoJSONL) using Fiona with spatial bbox filtering for memory-efficient geospatial ingestion.
• Engineered features via PySpark / Spark SQL — computed Flood_Height_Ratio and Flood_House_Damage_Ratio (sensor-confirmed) from building surface area projections (EPSG:3857) and flood depth readings; derived Damage_Classification using ratio thresholds (0.4 / 0.8).
• Constructed a predictive CatBoost GBDT multi-class classifier (iterations=1000, depth=6, lr=0.1, MultiClass loss, TotalF1 eval metric) for ECT allocation (₱0 / ₱5,000 / ₱10,000); achieved 99.99% accuracy, 1.00 weighted F1-score, 1.00 precision, and 1.00 recall on 922K real building footprints.
• Produced a confusion matrix confirming near-perfect classification across all three ECT classes — 138,176 / 19,279 / 27,100 correct predictions for ₱0 / ₱5,000 / ₱10,000 respectively.
• Generated a Seaborn correlation heatmap (YlGn colormap, zero-variance column filtering) to visualize feature-target relationships. Key Pearson correlation coefficients with Target_ECT_Amount: Flood_House_Damage_Ratio (r = 0.9293), Flood_Height_Ratio (r = 0.4892), Sensor_Flood_Depth_Meters (r = 0.3379), Flood_Depth_Meters (r = 0.2777), House_Surface_Area_Sqm (r = −0.0751). Notable inter-feature correlations: Flood_Depth_Meters ↔ Sensor_Flood_Depth_Meters (r = 0.9252), Flood_Height_Ratio ↔ Flood_Depth_Meters (r = 0.8222), Flood_House_Damage_Ratio ↔ Flood_Height_Ratio (r = 0.5961) — validating the engineered ratio features as primary drivers of ECT classification.
• Loaded the trained .bin model into a Django REST API with JSON-based inference endpoints and real-time model loading for automated ECT computation fed from SQLite3 database.
• Implemented Leaflet.js + OpenStreetMap GIS mapping with dynamic JavaScript logic, hazard visualization layers, and optimized Django database interactions.

---

References

1. DSWD (2025). Emergency Cash Transfer (ECT) Program Guidelines. https://www.dswd.gov.ph
2. Philippine Statistics Authority (2025). NICTHS 2024 Survey Findings. https://psa.gov.ph/nichths-2024
3. NDRRMC (2025). SitRep No. 5: Typhoon Uwan (Fung-wong). https://ndrrmc.gov.ph
4. NDRRMC (2025). SitRep No. 3: Typhoon Tino (Kalmaegi). https://ndrrmc.gov.ph
5. DBM (2025). National Calamity Fund Utilization Report. https://www.dbm.gov.ph/calamity-fund-2025
6. Office of Civil Defense (2025). PDNA Preliminary Report: Typhoon Uwan. https://ocd.gov.ph/pdna-uwan-2025
7. PAGASA (2025). Typhoon Season Summary. https://www.pagasa.dost.gov.ph/typhoon-report-2025
8. Republic Act No. 10639. Free Mobile Disaster Alerts Act. https://www.officialgazette.gov.ph/2014/06/20/republic-act-no-10639/
9. WHO Philippines (2025). SPEED System Performance Report. https://www.who.int/philippines/speed-2025
10. Smart Communications (2025). Emergency Alert System Report. (Internal Report — No public link)
11. Globe Telecom (2025). Disaster Response SMS Gateway Specifications. https://www.globe.com.ph/disaster-response
12. Litslink (2025). Cost Estimation for GIS-Based Disaster Aid Platforms. https://litslink.com/cost-estimate-gis-app
13. DOST-ASTI PhilSensors — Real-Time Water Level Monitoring (used by scrape_waterlevel.py) https://philsensors.asti.dost.gov.ph/site/waterlevel
14. Microsoft Global ML Building Footprints — Philippines GeoJSONL (used by scrape_households.py) https://github.com/microsoft/GlobalMLBuildingFootprints Direct download: https://minedbuildings.z5.web.core.windows.net/legacy/southeast-asia/philippines.geojsonl.zip