evo-lake-factor-attribution

Uses Random Forest feature importances to attribute water temperature changes to physical driver categories.

Category Mapping

Category	Variables	Description
Heat	AirTempLake, Shortwave, Longwave	Thermodynamic forcing
Flow	Precip, Outflow, Inflow	Hydrological forcing
Wind	WindSpeedLake	Mechanical forcing
Human	DevelopedArea, AgricultureArea	Anthropogenic forcing

Usage

import sys
sys.path.insert(0, '/app/environment/skills/evo-lake-factor-attribution/scripts')
from utils import train_random_forest, compute_feature_importances, map_features_to_categories, get_dominant_category, save_dominant_factor

# Train RF model
model = train_random_forest(X, y)

# Get feature importances
importance_df = compute_feature_importances(model, feature_names)

# Map to categories and aggregate
category_contributions = map_features_to_categories(importance_df, category_map)

# Get dominant category
category_name, contribution = get_dominant_category(category_contributions)

# Save output (columns: variable, contribution)
save_dominant_factor(category_name, contribution, '/root/output/dominant_factor.csv')

Key Functions

• train_random_forest(X, y) — fits RF regressor with random_state=42
• compute_feature_importances(model, feature_names) — extracts importances as percentages
• map_features_to_categories(importance_df, category_map) — aggregates by category
• get_dominant_category(category_contributions) — returns (name, contribution_decimal)
• save_dominant_factor(name, contribution, path) — writes single-row CSV

Output Format

dominant_factor.csv:

variable,contribution
Heat,0.45

Note: contribution is in decimal form (0-1 range), NOT percentage.

Import Pattern (avoiding naming conflicts)

When using multiple skills that each have utils.py, use importlib to avoid conflicts:

import importlib.util

def load_module(name, path):
    spec = importlib.util.spec_from_file_location(name, path)
    mod = importlib.util.module_from_spec(spec)
    spec.loader.exec_module(mod)
    return mod

data_utils = load_module('data_utils', '/app/environment/skills/evo-lake-data-pipeline/scripts/utils.py')
trend_utils = load_module('trend_utils', '/app/environment/skills/evo-lake-trend-analysis/scripts/utils.py')
factor_utils = load_module('factor_utils', '/app/environment/skills/evo-lake-factor-attribution/scripts/utils.py')