Africa Development Risk Index Model
AutoScientist Challenge 2026 | Marketing Category
Author: Hussein Adeiza (mabera)
Role: Licensed Environmental Health Officer, Abuja Nigeria
Base Model: Gemma 3 1B (it)
Fine-tuned with: AutoScientist by Adaption Labs
Model Description
This is a LoRA adapter fine-tuned to explain and pitch the Africa Development
Risk Index, a composite scoring system combining Health Risk, Economic Risk
and Legal Complexity into one unified view for African states and countries.
It answers product positioning questions for 8 distinct decision-maker audiences.
The Product
The Africa Development Risk Index synthesizes three models already built
in this challenge:
- Healthcare: Nigeria WASH Risk Model + Nigeria Malaria Health Model
- Finance: Nigeria Poverty Prediction Model
- Legal: Africa Environmental Law Model
Instead of three disconnected reports, decision-makers get one unified
composite score per Nigerian state or African country.
Training Data
- Source: Original product positioning Q&A authored by the project creator
- Dataset: 14 Q&A pairs across 8 audiences, expanded via Adaptive Data
- Languages: English, Hausa, Yoruba, French
- Quality improvement: 50.0% (Grade C → A)
- Kaggle: https://www.kaggle.com/datasets/yunusahusseinadeiza/africa-development-risk-index-marketing-dataset
Training Metrics
- Win rate: 57% adapted vs 43% base model
- Base model: google/gemma-3-1b-it
- Method: LoRA — House Special + Reasoning Traces + Hallucination mitigation
- Dataset quality: 6.0 → 9.0 (+50.0% improvement, Grade A)
Target Audiences
NGO Program Directors, Impact Investors, Government Policy Analysts,
Environmental Compliance Officers, Public Health Researchers, Startup
Founders, Journalists, Microfinance Institutions
Why This Matters
NGOs, investors, governments and startups working in Africa typically
consult fragmented, siloed data sources, health ministries, finance
ministries, environmental regulators, with no single view connecting
them. This model explains how a unified cross-domain index helps each
of these audiences make faster, better-informed decisions.
Credits
Powered by Adaptive Data — Adaption Labs
AutoScientist Challenge 2026 | Marketing Category
Built on top of: WASH, Malaria, Poverty and Africa Environmental Law models
