GIGA School Priority Backend

Backend pipeline to compute school-level composite priority indices based on multi-layer climate, social, and accessibility indicators.

This repository contains the full modular implementation of the school prioritisation framework developed within the GIGA Mozambique pilot.

---

Overview

The system computes a composite PRIORITY_IDX (0–100) for each school by combining:

• Climate vulnerability layers
• Social vulnerability layers
• Accessibility and isolation layers

The weighting structure is hierarchical:

1. Top-level weights between:

   • Climate
   • Social
   • Access

2. Internal subgroup weights within each category.

All scenario combinations are automatically generated using a structured preference system (High = 3, Medium = 2, Low = 1).

---

Repository Structure

giga-school-priority/
│
├─ src/giga_priority/         # Core Python package
│  ├─ scaling.py
│  ├─ indices.py
│  ├─ scenarios.py
│  ├─ export.py
│  └─ utils.py
│
├─ scripts/
│  ├─ run_export_school_priority.py
│  └─ inspect_scenarios.py
│
├─ data/
│  ├─ raw/
│  ├─ processed/
│  └─ outputs/
│
├─ requirements.txt
├─ pyproject.toml
└─ README.md

---

Installation

1️⃣ Clone the repository

git clone https://gitlab.com/giga-unicef/backend.git
cd backend

2️⃣ Create virtual environment

python -m venv .venv
.venv\Scripts\activate

3️⃣ Install dependencies

pip install -r requirements.txt
pip install -e .

---

Input Data

The pipeline expects a master CSV file:

data/processed/schools_master.csv

This file must contain:

Required Metadata Columns

• school_id_giga
• school_name
• education_level
• latitude
• longitude
• x_utm
• y_utm
• admin2_id_moai
• electricity_availability
• connectivity

Required Indicator Columns

All base columns required by:

• Climate index builder
• Social index builder
• Access index builder

These should already be preprocessed and cleaned.

---

Connectivity Filtering

The script automatically filters:

connectivity == "No"

Only non-connected schools are processed.

---

Running the Pipeline

Test Scenario Generation

python scripts/inspect_scenarios.py

Expected output:

N escenarios: 205

---

Run Full Export

python scripts/run_export_school_priority.py

This will:

• Load master dataset (~13,000 schools)
• Filter non-connected (~9,000)
• Generate 205 scenarios
• Compute indices for each
• Export results

---

Output

Results are written to:

data/outputs/

Two folders are created:

school_priority_csv/
school_priority_geojson/

For each scenario, the following files are generated:

schools_priority_index-cliX_socY_accZ-....csv
schools_priority_index-cliX_socY_accZ-....geojson

Each file contains:

• All original school metadata
• PRIORITY_IDX (integer 0–100)

GeoJSON is exported in:

EPSG:4326

---

PRIORITY_IDX Definition

The priority index is:

1. Built from:

   • Climate index
   • Social index
   • Access index

2. Combined using scenario weights

3. Robustly normalized (quantile-based)

4. Scaled to integer range:

0 – 100

---

Scenario Logic

Each scenario defines:

• Top-level preference:
  • High (3)
  • Medium (2)
  • Low (1)

Converted into normalized weights:

w_i = score_i / sum(scores)

Internal subgroup combinations are generated automatically.

Total scenarios (default configuration):

205

---

Reproducibility

This repository is fully reproducible:

• No hardcoded paths
• Modular structure
• Explicit scenario generation
• Version-controlled code

Large datasets are excluded via .gitignore.

---

Development Notes

• Built for Python 3.10 / 3.11
• Uses geopandas for spatial export
• Designed to integrate later with:
  • Greedy optimisation
  • MIP refinement
  • Tower placement modules

---

License

MIT — Copyright © 2025–2026 MOAI Analytics · GIGA UNICEF–ITU