#!/usr/bin/env python3 """ Test: Green Energy Transition — GreenTransition.lean ===================================================== Tests five propositions from the Lean-verified GreenTransition module against energy data from learning curve literature and OWID. Model: Y = [alpha*E_c^rho + (1-alpha)*E_d^rho]^{1/rho} Clean cost: p_c(Q) = p_0 * Q^{-beta} (Wright's Law) Geometric dimension: beta = d * alpha_0 Five sub-tests: 1. Three-group geometric dimension ordering (d=1 < energy < d=2) 2. Wright's Law fit for solar PV (R^2 > 0.90, negative slope) 3. Adoption discontinuity (structural break near cost parity) 4. Three-group ordering with statistical tests (Kruskal-Wallis) 5. Carbon tax acceleration (higher carbon price -> more renewables) Data: learning_curve_literature.csv, OWID energy data. """ import numpy as np import pandas as pd from scipy import stats from datetime import datetime import sys, os sys.path.insert(0, os.path.join(os.path.dirname(__file__), '..', '..')) from common.data import load_compiled from common.stats import verdict, log_result from common.plotting import setup_figure, setup_subplots, save_figure # ── Data loading ────────────────────────────────────────────────── OWID_URL = ("https://raw.githubusercontent.com/owid/energy-data/" "master/owid-energy-data.csv") _owid_cache = None def load_owid(): """Load OWID energy data. Cache in memory after first load.""" global _owid_cache if _owid_cache is not None: return _owid_cache from common.config import COMPILED_DIR cache_path = COMPILED_DIR / 'owid_energy_data.csv' if cache_path.exists(): age_days = (datetime.now().timestamp() - os.path.getmtime(cache_path)) / 86400 if age_days < 30: _owid_cache = pd.read_csv(cache_path, low_memory=False) return _owid_cache try: import requests r = requests.get(OWID_URL, timeout=60) r.raise_for_status() cache_path.parent.mkdir(parents=True, exist_ok=True) with open(cache_path, 'wb') as f: f.write(r.content) _owid_cache = pd.read_csv(cache_path, low_memory=False) return _owid_cache except Exception: return None def load_learning_curves(): """Load learning curve literature data.""" try: return load_compiled('learning_curve_literature') except FileNotFoundError: return None # ── Geometric dimension classification ──────────────────────────── D2_KEYWORDS = ['NAND Flash', 'HBM', 'DRAM', 'Intel micro'] ENERGY_KEYWORDS = ['Solar PV', 'Lithium-ion'] D1_KEYWORDS = ['Wind', 'Ford Model T', 'ATM', 'Electrolyzer', 'Credit card', 'Airbus', 'Toyota', 'M-Pesa'] def classify_dimension(row): """Classify technology by geometric dimension from product name.""" product = str(row.get('product', '')) industry = str(row.get('industry', '')) combined = product + ' ' + industry for kw in D2_KEYWORDS: if kw.lower() in combined.lower(): return 2 for kw in ENERGY_KEYWORDS: if kw.lower() in combined.lower(): return 1.5 for kw in D1_KEYWORDS: if kw.lower() in combined.lower(): return 1 return None def prepare_learning_data(): """Load and classify learning curve data.""" lc = load_learning_curves() if lc is None: return None, None, None, None lc = lc[lc['alpha'].notna() & (lc['alpha'] > 0) & (lc['alpha'] < 1)].copy() lc['d'] = lc.apply(classify_dimension, axis=1) lc = lc[lc['d'].notna()].copy() lc = lc[~lc['industry'].str.contains('Meta', na=False)] if 'notes' in lc.columns: lc = lc[~lc['notes'].str.contains('INFLATED|Implausibly', na=False, regex=True)] d2 = lc[lc['d'] == 2]['alpha'].values d_energy = lc[lc['d'] == 1.5]['alpha'].values d1 = lc[lc['d'] == 1]['alpha'].values return lc, d2, d_energy, d1 # ── Sub-tests ───────────────────────────────────────────────────── def test1_geometric_dimension(): """Test 1: Three-group ordering alpha(d=1) < alpha(energy) < alpha(d=2).""" lc, d2, d_energy, d1 = prepare_learning_data() if lc is None: return log_result('green_transition_geometric_ordering', 0.0, 1.0, 'SKIPPED', description="Tests three-group ordering of learning rates by geometric dimension: alpha(d=1) < alpha(energy) < alpha(d=2). Uses Mann-Whitney U test on learning_curve_literature.csv data classified by technology type.") if len(d2) < 2 or len(d1) < 2: return log_result('green_transition_geometric_ordering', 0.0, 1.0, 'SKIPPED', description="Tests three-group ordering of learning rates by geometric dimension: alpha(d=1) < alpha(energy) < alpha(d=2). Uses Mann-Whitney U test on learning_curve_literature.csv data classified by technology type.") # Key test: d2 > d1 _, p_2_gt_1 = stats.mannwhitneyu(d2, d1, alternative='greater') ratio_2_1 = d2.mean() / d1.mean() if d1.mean() > 0 else float('inf') # Check three-group ordering means_ordered = True p_energy = 1.0 if len(d_energy) >= 2: means_ordered = d1.mean() < d_energy.mean() < d2.mean() _, p_energy = stats.mannwhitneyu(d_energy, d1, alternative='greater') stat = ratio_2_1 p_val = p_2_gt_1 if means_ordered and p_2_gt_1 < 0.05: sign_correct = True else: sign_correct = False verd = verdict(p_val, sign_correct) if means_ordered and p_2_gt_1 < 0.05 and len(d_energy) >= 2 and p_energy >= 0.10: verd = 'AMBIGUOUS' return log_result( 'green_transition_geometric_ordering', stat, p_val, verd, {'d2_mean': d2.mean(), 'd1_mean': d1.mean(), 'd_energy_mean': d_energy.mean() if len(d_energy) > 0 else None, 'ratio_2_1': ratio_2_1}, description="Tests three-group ordering of learning rates by geometric dimension: alpha(d=1) < alpha(energy) < alpha(d=2). Uses Mann-Whitney U test (one-sided) on learning_curve_literature.csv data classified by technology type." ) def test2_wrights_law(): """Test 2: Wright's Law fit for solar PV.""" solar_data = pd.DataFrame({ 'year': [1976, 1980, 1985, 1990, 1995, 2000, 2005, 2008, 2010, 2012, 2014, 2016, 2018, 2020, 2022, 2023], 'cost_per_watt': [76.0, 30.0, 12.0, 8.0, 5.5, 4.0, 3.5, 2.5, 1.5, 0.7, 0.55, 0.35, 0.25, 0.22, 0.26, 0.22], 'cumulative_gw': [0.001, 0.01, 0.1, 0.5, 0.8, 1.5, 5.0, 16.0, 40.0, 100.0, 180.0, 300.0, 500.0, 750.0, 1200.0, 1500.0] }) log_cost = np.log(solar_data['cost_per_watt'].values) log_Q = np.log(solar_data['cumulative_gw'].values) slope, intercept, r_value, p_value, se = stats.linregress(log_Q, log_cost) beta_fit = -slope r2 = r_value ** 2 if r2 > 0.90 and slope < 0: sign_correct = True verd = verdict(p_value, sign_correct) elif r2 > 0.80 and slope < 0: verd = 'AMBIGUOUS' sign_correct = True else: sign_correct = False verd = verdict(p_value, sign_correct) return log_result( "green_transition_wrights_law", r2, p_value, verd, {'beta_fit': beta_fit, 'r_squared': r2, 'beta_literature': 0.23}, description="Tests Wright's Law for solar PV: log(cost) ~ log(cumulative capacity) should yield R^2 > 0.90 and a negative slope. Uses OLS on hardcoded solar PV cost-per-watt data from 1976-2023 against cumulative installed GW." ) def test3_adoption_discontinuity(): """Test 3: Structural break in solar adoption near cost parity.""" owid = load_owid() if owid is None: return log_result('green_transition_adoption_discontinuity', 0.0, 1.0, 'SKIPPED', description="Tests for a structural break in solar adoption near cost parity (2008-2018). Uses a Chow test on log(annual solar additions) from OWID world energy data.") world = owid[owid['country'] == 'World'].copy() if 'solar_electricity' not in world.columns: return log_result('green_transition_adoption_discontinuity', 0.0, 1.0, 'SKIPPED', description="Tests for a structural break in solar adoption near cost parity (2008-2018). Uses a Chow test on log(annual solar additions) from OWID world energy data.") world = world[['year', 'solar_electricity']].dropna().sort_values('year') world = world[world['year'] >= 1990] world['additions'] = world['solar_electricity'].diff() world = world[world['additions'].notna() & (world['additions'] > 0)] if len(world) < 10: return log_result('green_transition_adoption_discontinuity', 0.0, 1.0, 'SKIPPED', description="Tests for a structural break in solar adoption near cost parity (2008-2018). Uses a Chow test on log(annual solar additions) from OWID world energy data.") years = world['year'].values log_additions = np.log(world['additions'].values) n = len(years) try: F_crit = stats.f.ppf(0.95, 2, n - 4) except Exception: F_crit = 3.0 # Chow test at each candidate year x_full = years - years[0] slope_f, intercept_f, _, _, _ = stats.linregress(x_full, log_additions) ssr_full = np.sum((log_additions - (intercept_f + slope_f * x_full)) ** 2) expected_range = (2008, 2018) best_F = 0.0 best_year = None best_in_range_F = 0.0 best_in_range_year = None for split_idx in range(5, n - 5): split_year = years[split_idx] x1, y1 = x_full[:split_idx], log_additions[:split_idx] x2, y2 = x_full[split_idx:], log_additions[split_idx:] if len(x1) < 3 or len(x2) < 3: continue s1, i1, _, _, _ = stats.linregress(x1, y1) s2, i2, _, _, _ = stats.linregress(x2, y2) ssr_split = (np.sum((y1 - (i1 + s1 * x1)) ** 2) + np.sum((y2 - (i2 + s2 * x2)) ** 2)) k = 2 if ssr_split > 0: F = ((ssr_full - ssr_split) / k) / (ssr_split / (n - 2 * k)) if F > best_F: best_F = F best_year = split_year if (expected_range[0] <= split_year <= expected_range[1] and F > best_in_range_F): best_in_range_F = F best_in_range_year = split_year has_break = best_in_range_F > F_crit sign_correct = has_break if has_break: p_val = 1 - stats.f.cdf(best_in_range_F, 2, n - 4) verd = verdict(p_val, sign_correct) elif best_F > F_crit: p_val = 1 - stats.f.cdf(best_F, 2, n - 4) verd = 'AMBIGUOUS' else: p_val = 1 - stats.f.cdf(best_F, 2, n - 4) if best_F > 0 else 1.0 verd = 'AMBIGUOUS' return log_result( 'green_transition_adoption_discontinuity', best_in_range_F if has_break else best_F, p_val, verd, {'best_break_year': best_year, 'best_F': best_F, 'in_range_year': best_in_range_year, 'in_range_F': best_in_range_F, 'F_crit': F_crit}, description="Tests for a structural break in solar adoption near cost parity (2008-2018). Uses a Chow test scanning candidate break years in log(annual solar additions) from OWID data; the F-statistic must exceed the critical value within the expected range." ) def test4_three_group_ordering(): """Test 4: Three-group ordering with Kruskal-Wallis test.""" lc, d2, d_energy, d1 = prepare_learning_data() if lc is None or len(d2) < 2 or len(d1) < 3: return log_result('green_transition_three_group_ordering', 0.0, 1.0, 'SKIPPED', description="Tests that d=2 technologies (semiconductors) have significantly higher learning rates than d=1 technologies (linear manufacturing), with energy intermediate. Uses Welch t-test and Cohen's d effect size on learning_curve_literature.csv.") # Welch t-test: d=2 > d=1 t_stat, p_ttest = stats.ttest_ind(d2, d1, equal_var=False, alternative='greater') ratio_2_1 = d2.mean() / d1.mean() if d1.mean() > 0 else float('inf') # Effect size pooled_std = np.sqrt((d2.var() + d1.var()) / 2) cohens_d = (d2.mean() - d1.mean()) / pooled_std if pooled_std > 0 else 0 energy_intermediate = False if len(d_energy) >= 2: energy_intermediate = d1.mean() < d_energy.mean() < d2.mean() if p_ttest < 0.05 and 1.3 < ratio_2_1 < 3.0: if len(d_energy) >= 2 and energy_intermediate: sign_correct = True elif len(d_energy) < 2: sign_correct = True else: sign_correct = False else: sign_correct = False verd = verdict(p_ttest, sign_correct) return log_result( 'green_transition_three_group_ordering', t_stat, p_ttest, verd, {'ratio_2_1': ratio_2_1, 'cohens_d': cohens_d, 'energy_intermediate': energy_intermediate}, description="Tests that d=2 technologies (semiconductors) have significantly higher learning rates than d=1 technologies (linear manufacturing), with energy intermediate. Uses Welch t-test (one-sided) and Cohen's d effect size; expects ratio in [1.3, 3.0]." ) def test5_carbon_tax(): """Test 5: Higher carbon price -> higher renewable share.""" owid = load_owid() if owid is None: return log_result('green_transition_carbon_tax', 0.0, 1.0, 'SKIPPED', description="Tests whether higher carbon prices are associated with higher renewable energy shares across countries. Uses Spearman rank correlation between hardcoded carbon prices (USD/tCO2) and OWID renewable electricity share data.") ren_col = None for col in ['renewables_share_elec', 'renewables_share_energy']: if col in owid.columns: ren_col = col break if ren_col is None: return log_result('green_transition_carbon_tax', 0.0, 1.0, 'SKIPPED', description="Tests whether higher carbon prices are associated with higher renewable energy shares across countries. Uses Spearman rank correlation between hardcoded carbon prices (USD/tCO2) and OWID renewable electricity share data.") latest = owid.sort_values('year').groupby('country').last().reset_index() latest = latest[['country', 'year', ren_col]].dropna() rows = [] for country, price in carbon_prices.items(): match = latest[latest['country'] == country] if len(match) > 0: rows.append({ 'country': country, 'carbon_price': price, 'renewable_share': match.iloc[0][ren_col] }) df = pd.DataFrame(rows) if len(df) < 10: return log_result('green_transition_carbon_tax', 0.0, 1.0, 'SKIPPED', description="Tests whether higher carbon prices are associated with higher renewable energy shares across countries. Uses Spearman rank correlation between hardcoded carbon prices (USD/tCO2) and OWID renewable electricity share data.") spearman_r, spearman_p = stats.spearmanr(df['carbon_price'], df['renewable_share']) sign_correct = spearman_r > 0 verd = verdict(spearman_p, sign_correct) return log_result( 'green_transition_carbon_tax', spearman_r, spearman_p, verd, {'n_countries': len(df), 'spearman_r': spearman_r}, description="Tests whether higher carbon prices are associated with higher renewable energy shares across countries. Uses Spearman rank correlation between hardcoded carbon prices (USD/tCO2) and OWID renewable electricity share for 36 countries; prediction is positive correlation." ) # ── Main ────────────────────────────────────────────────────────── _cache = {} def main(): """Run all green transition tests. Returns list of result dicts.""" results = [] results.append(test1_geometric_dimension()) results.append(test2_wrights_law()) results.append(test3_adoption_discontinuity()) results.append(test4_three_group_ordering()) results.append(test5_carbon_tax()) _cache['results'] = results # Cache learning curve data for figures lc, d2, d_energy, d1 = prepare_learning_data() if lc is not None: _cache['d2'] = d2 _cache['d_energy'] = d_energy _cache['d1'] = d1 # Cache carbon tax data for figure owid = load_owid() if owid is not None: ren_col = None for col in ['renewables_share_elec', 'renewables_share_energy']: if col in owid.columns: ren_col = col break if ren_col: latest = owid.sort_values('year').groupby('country').last().reset_index() latest = latest[['country', 'year', ren_col]].dropna() rows = [] for country, price in carbon_prices.items(): match = latest[latest['country'] == country] if len(match) > 0: rows.append({'country': country, 'carbon_price': price, 'renewable_share': match.iloc[0][ren_col]}) if rows: _cache['carbon_tax_df'] = pd.DataFrame(rows) # Cache adoption discontinuity data for figure if owid is not None: world = owid[owid['country'] == 'World'].copy() if 'solar_electricity' in world.columns: world = world[['year', 'solar_electricity']].dropna().sort_values('year') world = world[world['year'] >= 1990] world['additions'] = world['solar_electricity'].diff() world = world[world['additions'].notna() & (world['additions'] > 0)] if len(world) >= 10: _cache['adoption_data'] = world return results # Carbon prices as module-level dict for cache access carbon_prices = { 'Sweden': 130, 'Switzerland': 120, 'Norway': 90, 'Finland': 85, 'France': 52, 'Germany': 35, 'Netherlands': 85, 'Denmark': 85, 'Ireland': 48, 'United Kingdom': 55, 'Canada': 50, 'Spain': 85, 'Italy': 85, 'Austria': 35, 'Portugal': 85, 'Poland': 85, 'Belgium': 85, 'Japan': 3, 'South Korea': 18, 'China': 10, 'Mexico': 4, 'South Africa': 9, 'Colombia': 5, 'Chile': 5, 'United States': 0, 'Australia': 0, 'India': 0, 'Brazil': 0, 'Indonesia': 0, 'Russia': 0, 'Saudi Arabia': 0, 'Nigeria': 0, 'Argentina': 0, 'Turkey': 0, 'Thailand': 0, 'Vietnam': 0, } def figures(): """Generate green transition figures: one per subtest.""" if not _cache: main() paths = [] d2 = _cache.get('d2') d_energy = _cache.get('d_energy') d1 = _cache.get('d1') # Figure 1: Geometric dimension boxplot (covers geometric_ordering + three_group_ordering) if d2 is not None and d1 is not None and len(d2) >= 2 and len(d1) >= 2: fig, ax = setup_figure(10, 5) groups = [d1] labels = ['d=1\n(linear)'] colors = ['lightsalmon'] positions = [1] if d_energy is not None and len(d_energy) > 0: groups.append(d_energy) labels.append('d~1.5\n(energy)') colors.append('lightgreen') positions.append(2) groups.append(d2) labels.append('d=2\n(semiconductor)') colors.append('lightskyblue') positions.append(3 if d_energy is not None and len(d_energy) > 0 else 2) bp = ax.boxplot(groups, positions=positions, widths=0.5, patch_artist=True) for box, c in zip(bp['boxes'], colors): box.set_facecolor(c) scatter_colors = ['red', 'green', 'blue'] if len(groups) == 3 else ['red', 'blue'] for grp, pos, sc in zip(groups, positions, scatter_colors): ax.scatter(pos * np.ones(len(grp)), grp, alpha=0.7, color=sc, zorder=5) ax.axhline(0.12, color='red', linestyle='--', alpha=0.5, label='Predicted d=1 (0.12)') ax.axhline(0.24, color='blue', linestyle='--', alpha=0.5, label='Predicted d=2 (0.24)') ax.set_xticks(positions) ax.set_xticklabels(labels) ax.set_ylabel('Learning rate alpha') ax.set_title('Geometric Dimension: Three-Group Ordering') ax.legend(fontsize=9) # Get test results for annotation results = _cache.get('results', []) r1 = next((r for r in results if r['name'] == 'green_transition_geometric_ordering'), None) r4 = next((r for r in results if r['name'] == 'green_transition_three_group_ordering'), None) ann_parts = [] if r1: ann_parts.append(f"Mann-Whitney p={r1['p_value']:.4f}") if r4: ann_parts.append(f"Welch t p={r4['p_value']:.3f}, Cohen's d={r4.get('cohens_d', 0):.2f}") if ann_parts: ax.text(0.02, 0.98, '\n'.join(ann_parts), transform=ax.transAxes, va='top', ha='left', fontsize=8, bbox=dict(boxstyle='round', fc='wheat', alpha=0.5)) paths.extend(save_figure(fig, 'green_geometric_dimension', subfolder='macro')) # Figure 2: Wright's Law solar PV (2 panels) solar_data = pd.DataFrame({ 'year': [1976, 1980, 1985, 1990, 1995, 2000, 2005, 2008, 2010, 2012, 2014, 2016, 2018, 2020, 2022, 2023], 'cost_per_watt': [76.0, 30.0, 12.0, 8.0, 5.5, 4.0, 3.5, 2.5, 1.5, 0.7, 0.55, 0.35, 0.25, 0.22, 0.26, 0.22], 'cumulative_gw': [0.001, 0.01, 0.1, 0.5, 0.8, 1.5, 5.0, 16.0, 40.0, 100.0, 180.0, 300.0, 500.0, 750.0, 1200.0, 1500.0] }) log_cost = np.log(solar_data['cost_per_watt'].values) log_Q = np.log(solar_data['cumulative_gw'].values) slope, intercept, r_value, _, _ = stats.linregress(log_Q, log_cost) beta_fit = -slope r2 = r_value ** 2 fig, axes = setup_subplots(1, 2, width=14, height=5) axes[0].scatter(log_Q, log_cost, color='orange', s=50, zorder=5) Q_fit = np.linspace(log_Q.min(), log_Q.max(), 100) axes[0].plot(Q_fit, intercept + slope * Q_fit, 'r-', label=f'OLS: beta = {beta_fit:.3f} (R2 = {r2:.3f})') axes[0].set_xlabel('log(Cumulative Capacity, GW)') axes[0].set_ylabel('log(Module Cost, $/W)') axes[0].set_title("Wright's Law: Solar PV") axes[0].legend(fontsize=9) axes[0].grid(True, alpha=0.3) axes[1].semilogy(solar_data['year'], solar_data['cost_per_watt'], 'o-', color='orange', markersize=6) axes[1].axhline(1.0, color='green', linestyle='--', alpha=0.7, label='Grid parity ($1/W)') axes[1].set_xlabel('Year') axes[1].set_ylabel('Module Cost ($/W, log scale)') axes[1].set_title('Solar PV Cost Trajectory') axes[1].legend() axes[1].grid(True, alpha=0.3) fig.suptitle("Wright's Law Solar PV", fontsize=14, y=1.02) paths.extend(save_figure(fig, 'green_wrights_law', subfolder='macro')) # Figure 3: Adoption discontinuity (Chow test on solar additions) adoption = _cache.get('adoption_data') if adoption is not None: years = adoption['year'].values log_additions = np.log(adoption['additions'].values) n = len(years) x_full = years - years[0] # Reproduce Chow test to find best break slope_f, intercept_f, _, _, _ = stats.linregress(x_full, log_additions) ssr_full = np.sum((log_additions - (intercept_f + slope_f * x_full)) ** 2) best_year = None best_F = 0 for split_idx in range(5, n - 5): x1, y1 = x_full[:split_idx], log_additions[:split_idx] x2, y2 = x_full[split_idx:], log_additions[split_idx:] if len(x1) < 3 or len(x2) < 3: continue s1, i1, _, _, _ = stats.linregress(x1, y1) s2, i2, _, _, _ = stats.linregress(x2, y2) ssr_s = np.sum((y1 - (i1 + s1 * x1))**2) + np.sum((y2 - (i2 + s2 * x2))**2) if ssr_s > 0: F = ((ssr_full - ssr_s) / 2) / (ssr_s / (n - 4)) if 2008 <= years[split_idx] <= 2018 and F > best_F: best_F = F best_year = years[split_idx] fig, ax = setup_figure(10, 6) ax.scatter(years, log_additions, color='darkorange', s=40, zorder=5, label='log(annual additions)') ax.plot(years, intercept_f + slope_f * x_full, 'b--', alpha=0.5, label='Full-sample trend') if best_year is not None: idx = np.where(years == best_year)[0][0] x1, y1 = x_full[:idx], log_additions[:idx] x2, y2 = x_full[idx:], log_additions[idx:] s1, i1, _, _, _ = stats.linregress(x1, y1) s2, i2, _, _, _ = stats.linregress(x2, y2) ax.plot(years[:idx], i1 + s1 * x1, 'r-', linewidth=2, label=f'Pre-{best_year}') ax.plot(years[idx:], i2 + s2 * x2, 'g-', linewidth=2, label=f'Post-{best_year}') ax.axvline(best_year, color='gray', linestyle=':', alpha=0.7) F_crit = stats.f.ppf(0.95, 2, n - 4) ax.text(0.02, 0.02, f'Break year: {best_year}\nF={best_F:.1f} (crit={F_crit:.1f})', transform=ax.transAxes, va='bottom', fontsize=9, bbox=dict(boxstyle='round', fc='lightyellow', alpha=0.8)) ax.axvspan(2008, 2018, alpha=0.1, color='green', label='Expected parity range') ax.set_xlabel('Year') ax.set_ylabel('log(Solar Electricity Additions, TWh)') ax.set_title('Adoption Discontinuity: Structural Break in Solar Growth') ax.legend(fontsize=8, loc='upper left') ax.grid(True, alpha=0.3) paths.extend(save_figure(fig, 'green_adoption_discontinuity', subfolder='macro')) # Figure 4: Carbon tax scatter ct_df = _cache.get('carbon_tax_df') if ct_df is not None and len(ct_df) >= 10: fig, ax = setup_figure(10, 6) ax.scatter(ct_df['carbon_price'], ct_df['renewable_share'], s=50, alpha=0.7, color='teal') for _, row in ct_df.iterrows(): ax.annotate(row['country'], (row['carbon_price'], row['renewable_share']), fontsize=6, alpha=0.7, xytext=(3, 3), textcoords='offset points') sr, sp = stats.spearmanr(ct_df['carbon_price'], ct_df['renewable_share']) ax.set_xlabel('Carbon Price (USD/tCO2)') ax.set_ylabel('Renewable Share of Electricity (%)') ax.set_title(f'Carbon Tax vs Renewable Share (N={len(ct_df)})\nSpearman r={sr:.3f}, p={sp:.4f}') ax.grid(True, alpha=0.3) paths.extend(save_figure(fig, 'green_carbon_tax', subfolder='macro')) return paths if __name__ == '__main__': results = main() for r in results: print(f" {r['name']}: {r['verdict']} " f"(stat={r['statistic']:.4f}, p={r['p_value']:.4f})")

p value	2.840e-4
statistic	0.5701
spearman r	0.5701
n countries	36

Green Transition Carbon Tax

Description

Figures

Statistics

Script

▸ Source Code

Reproduction Steps

Data Sources