Merge pull request #12 from utorque/claude/add-disrupted-data-method-01HLMfF2mQgxyvpFv42KQ86t (6af7af3f) · Commits · Barthelet Thibault / horloml

.gitignore

+1 −0

Original line number	Diff line number	Diff line
		@@ -3,6 +3,7 @@
		/HorloML_EDU_Main/web_app/flask_session
		/supply_chain_sim.egg-info
		/forecast_app/data
		/data
		/venv
		/__pycache__
		*.egg

forecast_app/app.py

+67 −10

Original line number	Diff line number	Diff line
		@@ -34,13 +34,15 @@ app.config['SESSION_TYPE'] = 'filesystem'
		# Load dataset
		DATA_DIR = os.path.join(os.path.dirname(__file__), 'data')
		TRAINING_DATA_PATH = os.path.join(DATA_DIR, 'sim2_supply_chain_data_training.json')
		TEST_DATA_PATH = os.path.join(DATA_DIR, 'sim2_supply_chain_data_test.json')
		TEST_DATA_PATH_NORMAL = os.path.join(DATA_DIR, 'sim2_supply_chain_data_test.json')
		TEST_DATA_PATH_EXTERNAL = os.path.join(DATA_DIR, 'sim2_data_test_external_events.json')

		# TiRex model (loaded lazily on first request)
		_tirex_model = None

		# AI activation state (disabled by default)
		_ai_enabled = False
		# Teacher parameters
		_ai_enabled = True
		_simulation_type = 'normal' # 'normal' or 'external_event'
		AI_PASSWORD = os.environ.get('AI_PASSWORD', 'HorloML-AI')


		@@ -59,8 +61,23 @@ def load_training_data():


		def load_test_data():
		"""Load the year 11 test data (ground truth)"""
		with open(TEST_DATA_PATH, 'r') as f:
		"""Load the year 11 test data (ground truth) based on simulation type"""
		global _simulation_type

		if _simulation_type == 'external_event':
		# Load external events data
		if os.path.exists(TEST_DATA_PATH_EXTERNAL):
		with open(TEST_DATA_PATH_EXTERNAL, 'r') as f:
		data = json.load(f)
		# External events file has different structure: {'metadata': ..., 'test_data': [...]}
		return data.get('test_data', data)
		else:
		# Fallback to normal if external file doesn't exist
		with open(TEST_DATA_PATH_NORMAL, 'r') as f:
		return json.load(f)
		else:
		# Load normal test data
		with open(TEST_DATA_PATH_NORMAL, 'r') as f:
		return json.load(f)


		@@ -407,7 +424,7 @@ def download_excel():
		headers = ['Year', 'Month', 'Date']
		for watch in watches:
		headers.extend([
		f"{watch['name']} - Demand",
		f"{watch['name']} - Units Sold",
		f"{watch['name']} - Revenue",
		f"{watch['name']} - Profit"
		])
		@@ -445,7 +462,7 @@ def download_excel():
		yearly_headers = ['Year']
		for watch in watches:
		yearly_headers.extend([
		f"{watch['name']} - Total Demand",
		f"{watch['name']} - Total Units Sold",
		f"{watch['name']} - Total Revenue",
		f"{watch['name']} - Total Profit"
		])
		@@ -594,16 +611,56 @@ def results():
		test_data=test_data)


		@app.route('/api/teacher_params', methods=['GET'])
		def get_teacher_params():
		"""Get current teacher parameters (AI status and simulation type)"""
		global _ai_enabled, _simulation_type
		return jsonify({
		'ai_enabled': _ai_enabled,
		'simulation_type': _simulation_type
		})


		@app.route('/api/teacher_params', methods=['POST'])
		def update_teacher_params():
		"""Update teacher parameters with password verification"""
		global _ai_enabled, _simulation_type

		data = request.get_json()
		password = data.get('password', '')

		if password != AI_PASSWORD:
		return jsonify({'success': False, 'error': 'Invalid password'}), 401

		# Update parameters if provided
		if 'ai_enabled' in data:
		_ai_enabled = data['ai_enabled']

		if 'simulation_type' in data:
		sim_type = data['simulation_type']
		if sim_type in ['normal', 'external_event']:
		_simulation_type = sim_type
		else:
		return jsonify({'success': False, 'error': 'Invalid simulation type'}), 400

		return jsonify({
		'success': True,
		'ai_enabled': _ai_enabled,
		'simulation_type': _simulation_type
		})


		# Legacy endpoints for backward compatibility
		@app.route('/api/ai_status', methods=['GET'])
		def ai_status():
		"""Get current AI activation status"""
		"""Get current AI activation status (legacy endpoint)"""
		global _ai_enabled
		return jsonify({'enabled': _ai_enabled})


		@app.route('/api/toggle_ai', methods=['POST'])
		def toggle_ai():
		"""Toggle AI activation with password verification"""
		"""Toggle AI activation with password verification (legacy endpoint)"""
		global _ai_enabled

		data = request.get_json()

forecast_app/smart_data_generator.py

+127 −1

Original line number	Diff line number	Diff line
		@@ -389,7 +389,8 @@ class SmartSupplyChainDataGenerator:
		demand = purchases[watch_id]

		# Production strategy: produce based on demand + safety stock
		production = int(demand * 1.05) # TODO : here maybe do a random value between .95 and 1.05 for noise ?
		# production = int(demand * 1.05) # TODO : here maybe do a random value between .95 and 1.05 for noise ?
		production = int(demand * np.random.uniform(0.95, 1.05))

		# Calculate financials
		watch_data = self._calculate_costs_and_revenue(
		@@ -439,6 +440,113 @@ class SmartSupplyChainDataGenerator:

		return dataset['historical_data'][start_idx:end_idx]

		def generate_changed_test_data(self, dataset: Dict, disruption_config: List[Dict],
		test_year: int = 11, filepath: str = 'data/data_test_external_events.json'):
		"""
		Generate test data with external disruption events applied

		Args:
		dataset: Full dataset to extract test data from
		disruption_config: List of disruption events, each with:
		- month_to_change: Month number (1-12) to apply disruption
		- demand_ratio: Multiplier for demand (e.g., 0.5 halves demand, 2.0 doubles it)
		- watch_ids: Optional list of watch IDs to affect (default: all watches)
		test_year: Which year to use as test data (default: 11)
		filepath: Where to save the modified test data

		Returns:
		Modified test data with disruptions applied

		Example:
		disruption_config = [
		{'month_to_change': 3, 'demand_ratio': 0.5}, # Halve demand in March
		{'month_to_change': 6, 'demand_ratio': 0.3, 'watch_ids': [1, 2]} # 70% drop in June for watches 1&2
		]
		"""
		# Get base test data
		test_data = self.get_test_data(dataset, test_year)

		# Create a deep copy to modify
		import copy
		modified_test_data = copy.deepcopy(test_data)

		# Track inventory for recalculating financials
		# Get ending inventory from last training month
		last_training_idx = (test_year - 1) * 12 - 1
		inventory = {}
		for watch in self.watches:
		watch_id = watch['id']
		# Find the watch data in the last training month
		last_month_data = dataset['historical_data'][last_training_idx]
		watch_data = next(w for w in last_month_data['watches'] if w['watch_id'] == watch_id)
		inventory[watch_id] = watch_data['inventory_end']

		# Apply disruptions month by month
		for month_idx, month_data in enumerate(modified_test_data):
		month_in_year = month_data['month']

		# Check if any disruptions apply to this month
		disruptions_this_month = [
		d for d in disruption_config
		if d['month_to_change'] == month_in_year
		]

		# Process each watch
		for watch_idx, watch in enumerate(self.watches):
		watch_id = watch['id']
		watch_data = month_data['watches'][watch_idx]

		# Get current demand
		original_demand = watch_data['demand']
		modified_demand = original_demand

		# Apply all applicable disruptions for this watch
		for disruption in disruptions_this_month:
		# Check if this disruption affects this watch
		affected_watches = disruption.get('watch_ids', None)
		if affected_watches is None or watch_id in affected_watches:
		modified_demand = int(modified_demand * disruption['demand_ratio'])

		# If demand changed, recalculate all financials
		if modified_demand != original_demand:
		production = watch_data['production'] # Keep same production decision

		# Recalculate financials with new demand
		new_financials = self._calculate_costs_and_revenue(
		watch, modified_demand, production, inventory[watch_id]
		)

		# Update watch data with new values
		watch_data.update(new_financials)
		watch_data['watch_id'] = watch['id']
		watch_data['watch_name'] = watch['name']
		watch_data['original_demand'] = original_demand # Keep track of original
		watch_data['disruption_applied'] = True

		# Update inventory for next month
		inventory[watch_id] = watch_data['inventory_end']

		# Add metadata about disruptions
		output_data = {
		'metadata': {
		'generated_date': datetime.now().isoformat(),
		'generator_type': 'smart_customer_simulation_with_disruptions',
		'base_year': test_year,
		'disruptions': disruption_config,
		'watches': self.watches
		},
		'test_data': modified_test_data
		}

		# Save to file
		import os
		os.makedirs(os.path.dirname(filepath), exist_ok=True)
		with open(filepath, 'w') as f:
		json.dump(output_data, f, indent=2)
		print(f"Modified test data with disruptions saved to {filepath}")

		return output_data


		def main():
		"""Generate and save the dataset"""
		@@ -473,6 +581,22 @@ def main():
		with open('data/sim2_supply_chain_data_test.json', 'w') as f:
		json.dump(test_data, f, indent=2)

		# Generate test data with external disruptions (example)
		print("\n" + "-" * 60)
		print("Generating test data with external disruptions...")
		print("-" * 60)

		disruption_config = [
		{'month_to_change': 3, 'demand_ratio': 0.4}, # Halve demand in March (supply chain crisis)
		{'month_to_change': 4, 'demand_ratio': 0.4}, # 50% increase in July (recovery surge)
		]

		disrupted_data = generator.generate_changed_test_data(
		full_dataset,
		disruption_config,
		filepath='data/sim2_data_test_external_events.json'
		)

		print("\n" + "=" * 60)
		print("Dataset Generation Complete!")
		print("=" * 60)
		@@ -534,3 +658,5 @@ def main():

		if __name__ == "__main__":
		main()

forecast_app/templates/base.html

+208 −55

File changed.

Preview size limit exceeded, changes collapsed.

forecast_app/templates/historical.html

+3 −3

Original line number	Diff line number	Diff line
		@@ -89,7 +89,7 @@
		{% for watch in watches %}
		<div class="column is-4">
		<div class="box">
		<h4 class="title is-5">{{ watch.name }} - Monthly Demand</h4>
		<h4 class="title is-5">{{ watch.name }} - Monthly Units</h4>
		<div class="chart-container" style="height: 300px;">
		<canvas id="chart_{{ watch.id }}"></canvas>
		</div>
		@@ -115,7 +115,7 @@
		<th></th>
		<th></th>
		{% for watch in watches %}
		<th class="has-text-right">Demand</th>
		<th class="has-text-right">Units</th>
		<th class="has-text-right">Revenue</th>
		{% endfor %}
		</tr>
		@@ -168,7 +168,7 @@ watches.forEach(watch => {
		data: {
		labels: labels,
		datasets: [{
		label: 'Monthly Demand',
		label: 'Monthly Units Sold',
		data: watchData,
		borderColor: watch.id === 1 ? '#667eea' : watch.id === 2 ? '#f093fb' : '#4facfe',
		backgroundColor: watch.id === 1 ? '#667eea33' : watch.id === 2 ? '#f093fb33' : '#4facfe33',