SAFSI Methodology

Comprehensive explanation of how the South African Food Security Index is calculated

The SAFSI employs a rigorous, data-driven methodology to transform raw data into meaningful scores. This page explains the data collection, normalization, weighting, and aggregation processes that produce the final index.

Methodology Overview

The SAFSI methodology follows a hierarchical approach to transform raw data into a comprehensive food security index:

  1. Data Collection: Raw data is gathered from multiple sources including government statistics, industry reports, surveys, and satellite imagery.
  2. Data Processing: Raw data is cleaned, validated, and transformed into consistent formats and units.
  3. Metric Calculation: Individual metrics are calculated from processed data using specific formulas for each indicator.
  4. Normalization: Metrics are standardized using Z-score or min-max normalization to create comparable scales (0-100).
  5. Weighting: Each metric is assigned a weight based on its importance to food security and data reliability.
  6. Aggregation: Weighted metrics are combined to create pillar scores, which are then weighted and combined into the final SAFSI score.

SAFSI Data Flow

How raw data flows through the system to create the composite index

Sample Code: SAFSI Calculation

Python code example showing how the SAFSI is calculated from normalized metrics

# Pillar Structure and Scoring Logic

# Pillar Definitions with Weights
pillars = {
    "Input & Production": {
        "weight": 0.20,
        "metrics": {
            "Crop Yield Index": 0.05,
            "Livestock Productivity Index": 0.05,
            "Input Cost Index": 0.03,
            "Input Chain Resilience Score": 0.03,
            "Labor Availability Index": 0.02,
            "Land & Resource Access Score": 0.02
        }
    },
    "Processing & Logistics": {
        "weight": 0.15,
        "metrics": {
            "Post-Harvest Loss Rate": 0.04,
            "Cold Chain Integrity Index": 0.03,
            "Processing Efficiency Index": 0.03,
            "Storage Utilization Score": 0.02,
            "Road & Port Logistics Score": 0.03
        }
    },
    "Access & Markets": {
        "weight": 0.20,
        "metrics": {
            "Food Affordability Index": 0.05,
            "Price Transmission Efficiency": 0.04,
            "Export Performance Index": 0.03,
            "Domestic Market Connectivity": 0.03,
            "Retail Accessibility Score": 0.03,
            "Consumer Sentiment Index": 0.02
        }
    },
    "Consumption & Nutrition": {
        "weight": 0.15,
        "metrics": {
            "Dietary Diversity Score": 0.05,
            "Food Safety Compliance": 0.04,
            "Consumer Satisfaction Index": 0.03,
            "Nutrition Outcomes (Stunting, etc.)": 0.03
        }
    },
    "Environmental Resilience": {
        "weight": 0.15,
        "metrics": {
            "Water Use Efficiency": 0.04,
            "GHG Emissions Intensity": 0.03,
            "Biodiversity Impact Score": 0.03,
            "Circular Economy Adoption": 0.03,
            "Rainfall/Climate Stress Index": 0.02
        }
    },
    "Socio-Institutional Stability": {
        "weight": 0.15,
        "metrics": {
            "Rural Employment & Income Index": 0.04,
            "Rural Crime & Safety Score": 0.03,
            "Land Equity & Reform Index": 0.03,
            "Agricultural Confidence Index": 0.03,
            "Policy & Governance Score (Delphi/AHP)": 0.02
        }
    }
}

# Normalize and score metrics between 0 and 100
def normalize_metric(value, min_val, max_val, invert=False):
    if max_val == min_val:
        return 50  # prevent division by zero
    score = ((value - min_val) / (max_val - min_val)) * 100
    return 100 - score if invert else score

# Compute index score from normalized metrics
def compute_pillar_score(pillar_metrics):
    total_score = 0
    for metric, data in pillar_metrics.items():
        norm_score = data['score']  # pre-normalized to 0–100 externally
        weight = data['weight']
        total_score += norm_score * weight
    return total_score

# Calculate full AFSI score
def compute_afsi_index(metrics_input):
    afsi_score = 0
    for pillar, config in pillars.items():
        pillar_score = 0
        metric_scores = {}
        for metric, weight in config['metrics'].items():
            score = metrics_input.get(metric, 50)  # default to 50 if missing
            metric_scores[metric] = {'score': score, 'weight': weight}
        pillar_score = compute_pillar_score(metric_scores)
        afsi_score += pillar_score * config['weight']
    return round(afsi_score, 2)

# Example usage
if __name__ == "__main__":
    # Placeholder input dictionary for demonstration (replace with real data)
    input_scores = {
        "Crop Yield Index": 70,
        "Livestock Productivity Index": 65,
        "Input Cost Index": 40,
        "Post-Harvest Loss Rate": 30,
        "Food Affordability Index": 60,
        "Dietary Diversity Score": 55,
        "Water Use Efficiency": 65,
        "Rural Crime & Safety Score": 50,
        "Land Equity & Reform Index": 45,
        "Policy & Governance Score (Delphi/AHP)": 70
    }
    index_result = compute_afsi_index(input_scores)
    print(f"AFSI Composite Score: {index_result}/100")