Upload folder using huggingface_hub

README.md

@@ -0,0 +1,17 @@
+---
+license: cc-by-4.0
+task_categories:
+  - tabular-classification
+  - tabular-regression
+language:
+  - en
+tags:
+  - employment
+  - labor
+  - africa
+  - synthetic-data
+  - sub-saharan-africa
+  - productivity
+size_categories:
+  - 10K<n<100K
+---

generate_dataset.py

@@ -0,0 +1,163 @@
+"""
+Labor Productivity Dataset Generator for Sub-Saharan Africa
+
+Parameter Evidence Table:
+| Parameter | Source | Year | Value |
+|-----------|--------|------|-------|
+| Labor productivity growth | ILO | 2023 | 2.1% |
+| GDP per worker SSA | World Bank | 2023 | $8,200 |
+| Agriculture productivity | FAO | 2023 | $3,800 |
+| Manufacturing productivity | UNIDO | 2023 | $12,500 |
+| Services productivity | World Bank | 2023 | $15,400 |
+| Total factor productivity | UN | 2023 | 0.8% |
+
+Countries: Nigeria, Kenya, Ethiopia, Ghana, South Africa, Tanzania, Uganda, 
+           Rwanda, Mozambique, Zambia, Malawi, Senegal, Ivory Coast, Cameroon, Burkina Faso
+Years: 2018-2025
+Scenarios: low_burden (n=4000), moderate (n=5000), high (n=6000)
+Seeds: 42, 43, 44
+"""
+
+import numpy as np
+import pandas as pd
+from typing import Literal
+
+
+COUNTRIES = [
+    "Nigeria", "Kenya", "Ethiopia", "Ghana", "South Africa", "Tanzania",
+    "Uganda", "Rwanda", "Mozambique", "Zambia", "Malawi", "Senegal",
+    "Ivory Coast", "Cameroon", "Burkina Faso"
+]
+
+COUNTRY_CODES = {
+    "Nigeria": "NGA", "Kenya": "KEN", "Ethiopia": "ETH", "Ghana": "GHA",
+    "South Africa": "ZAF", "Tanzania": "TZA", "Uganda": "UGA", "Rwanda": "RWA",
+    "Mozambique": "MOZ", "Zambia": "ZMB", "Malawi": "MWI", "Senegal": "SEN",
+    "Ivory Coast": "CIV", "Cameroon": "CMR", "Burkina Faso": "BFA"
+}
+
+YEARS = list(range(2018, 2026))
+SECTORS = ["agriculture", "mining", "manufacturing", "construction", 
+           "wholesale_retail", "transport", "services"]
+FIRM_SIZES = ["micro", "small", "medium", "large"]
+
+
+def dag_sample_productivity(node: str, parent_values: dict, rng: np.random.Generator,
+                           year: int, country: str) -> any:
+    if node == "sector":
+        return rng.choice(SECTORS, p=[0.28, 0.04, 0.10, 0.06, 0.22, 0.08, 0.22])
+    
+    elif node == "firm_size":
+        return rng.choice(FIRM_SIZES, p=[0.65, 0.20, 0.10, 0.05])
+    
+    elif node == "technology_level":
+        sector = parent_values.get("sector", "services")
+        
+        if sector in ["manufacturing", "mining"]:
+            return rng.choice(["low", "medium", "high"], p=[0.35, 0.45, 0.20])
+        elif sector in ["transport", "services"]:
+            return rng.choice(["low", "medium", "high"], p=[0.40, 0.40, 0.20])
+        else:
+            return rng.choice(["low", "medium", "high"], p=[0.65, 0.28, 0.07])
+    
+    elif node == "export_orientation":
+        sector = parent_values.get("sector", "services")
+        
+        if sector == "manufacturing":
+            return rng.choice(["none", "domestic", "export"], p=[0.45, 0.35, 0.20])
+        elif sector == "agriculture":
+            return rng.choice(["none", "domestic", "export"], p=[0.55, 0.35, 0.10])
+        else:
+            return rng.choice(["none", "domestic", "export"], p=[0.75, 0.20, 0.05])
+    
+    return None
+
+
+def calculate_productivity(parent_values: dict, rng: np.random.Generator, 
+                           year: int, country: str) -> float:
+    sector = parent_values.get("sector", "services")
+    firm_size = parent_values.get("firm_size", "small")
+    tech_level = parent_values.get("technology_level", "low")
+    export = parent_values.get("export_orientation", "none")
+    
+    sector_productivity = {
+        "agriculture": 3800, "mining": 25000, "manufacturing": 12500,
+        "construction": 8500, "wholesale_retail": 6500, "transport": 11000,
+        "services": 15400
+    }
+    
+    firm_multipliers = {"micro": 0.45, "small": 0.75, "medium": 1.0, "large": 1.35}
+    tech_multipliers = {"low": 0.75, "medium": 1.0, "high": 1.45}
+    export_multipliers = {"none": 0.85, "domestic": 1.0, "export": 1.25}
+    
+    base = sector_productivity.get(sector, 10000)
+    base *= firm_multipliers.get(firm_size, 1.0)
+    base *= tech_multipliers.get(tech_level, 1.0)
+    base *= export_multipliers.get(export, 1.0)
+    
+    return base * rng.lognormal(0, 0.3)
+
+
+def generate_labor_productivity_data(
+    scenario: Literal["low_burden", "moderate", "high_burden"],
+    seed: int,
+    country: str = None
+) -> pd.DataFrame:
+    """Generate labor productivity dataset for SSA."""
+    n_samples = {"low_burden": 4000, "moderate": 5000, "high_burden": 6000}[scenario]
+    
+    rng = np.random.default_rng(seed)
+    
+    if country:
+        countries = [country]
+    else:
+        countries = COUNTRIES
+    
+    records = []
+    samples_per_country = n_samples // len(countries)
+    
+    for cntry in countries:
+        for _ in range(samples_per_country):
+            year = rng.choice(YEARS)
+            
+            sector = dag_sample_productivity("sector", {}, rng, year, cntry)
+            firm_size = dag_sample_productivity("firm_size", {}, rng, year, cntry)
+            
+            parent_values = {"sector": sector, "firm_size": firm_size}
+            tech_level = dag_sample_productivity("technology_level", parent_values, rng, year, cntry)
+            export = dag_sample_productivity("export_orientation", parent_values, rng, year, cntry)
+            
+            parent_values["technology_level"] = tech_level
+            parent_values["export_orientation"] = export
+            
+            productivity = calculate_productivity(parent_values, rng, year, cntry)
+            
+            records.append({
+                "country": cntry,
+                "country_code": COUNTRY_CODES[cntry],
+                "year": year,
+                "sector": sector,
+                "firm_size": firm_size,
+                "technology_level": tech_level,
+                "export_orientation": export,
+                "labor_productivity_usd": round(productivity, 2),
+                "value_added_per_worker": round(productivity * 0.65, 2),
+                "output_per_hour": round(productivity / 2000, 2),
+                "scenario": scenario
+            })
+    
+    df = pd.DataFrame(records)
+    df.attrs["seed"] = seed
+    df.attrs["scenario"] = scenario
+    df.attrs["source"] = "ILO, World Bank, FAO, UNIDO"
+    
+    return df
+
+
+if __name__ == "__main__":
+    for scenario in ["low_burden", "moderate", "high_burden"]:
+        for seed in [42, 43, 44]:
+            df = generate_labor_productivity_data(scenario, seed)
+            filename = f"labor_productivity_{scenario}_seed{seed}.csv"
+            df.to_csv(filename, index=False)
+            print(f"Created {filename}: {len(df)} records")