Configuration Reference¶

Complete reference for all configuration options.

Configuration File¶

Main configuration in config/config.yaml:

# Full example with all options
data:
  nyc_taxi:
    path: "data/raw/nyc_taxi"
    zones: [1, 2, 3, 4, 5]
    time_range: "2023-01-01:2023-12-31"
    dataset_id: "new-york-city-taxi-fare-prediction"
  nasa_turbofan:
    path: "data/raw/nasa_turbofan"
    dataset_id: "nasa-turbofan-engine-degradation-simulation-data-set"
  processed_path: "data/processed"
  models_path: "data/models"
  outputs_path: "data/outputs"

fleet:
  num_vehicles: 50
  num_locations: 5
  simulation_seed: 42
  vehicle_capacity: 1
  max_travel_distance: 100

forecasting:
  model: "xgboost"
  horizon_days: 7
  features:
    - hour
    - day_of_week
    - month
  xgboost:
    n_estimators: 100
    max_depth: 5
    learning_rate: 0.1

optimization:
  solver: "ortools"
  stages:
    - min_cost_flow
  constraints_file: "config/constraints/fleet_constraints.json"

risk:
  model: "heuristic"
  heuristic_threshold: 0.8

api:
  host: "0.0.0.0"
  port: 8000
  log_level: "INFO"

logging:
  version: 1
  disable_existing_loggers: false

Data Configuration¶

NYC Taxi Data¶

Option	Type	Default	Description
`path`	string	`"data/raw/nyc_taxi"`	Storage path
`zones`	list[int]	`[1,2,3,4,5]`	Taxi zone IDs
`time_range`	string	`"2023-01-01:2023-12-31"`	Date range
`dataset_id`	string	-	Kaggle dataset ID

NASA Turbofan Data¶

Option	Type	Default	Description
`path`	string	`"data/raw/nasa_turbofan"`	Storage path
`dataset_id`	string	-	Kaggle dataset ID

Paths¶

Option	Type	Default	Description
`processed_path`	string	`"data/processed"`	Processed data
`models_path`	string	`"data/models"`	Model artifacts
`outputs_path`	string	`"data/outputs"`	Generated outputs

Fleet Configuration¶

Option	Type	Default	Description
`num_vehicles`	int	`50`	Total vehicles in fleet
`num_locations`	int	`5`	Number of service locations
`simulation_seed`	int	`42`	Random seed for reproducibility
`vehicle_capacity`	int	`1`	Passengers per vehicle
`max_travel_distance`	float	`100`	Max rebalancing distance (km)

Forecasting Configuration¶

General¶

Option	Type	Default	Description
`model`	string	`"xgboost"`	Model type: `xgboost`, `prophet`, `tft`
`horizon_days`	int	`7`	Forecast horizon
`features`	list[str]	`[hour, day_of_week, month]`	Feature columns

XGBoost Options¶

Option	Type	Default	Description
`n_estimators`	int	`100`	Number of trees
`max_depth`	int	`5`	Maximum tree depth
`learning_rate`	float	`0.1`	Learning rate
`min_child_weight`	int	`1`	Min child weight
`subsample`	float	`0.8`	Row subsampling
`colsample_bytree`	float	`0.8`	Column subsampling

Prophet Options (Phase 2)¶

Option	Type	Default	Description
`seasonality_mode`	string	`"multiplicative"`	Seasonality type
`changepoint_prior_scale`	float	`0.05`	Trend flexibility
`yearly_seasonality`	bool	`true`	Include yearly seasonality
`weekly_seasonality`	bool	`true`	Include weekly seasonality
`daily_seasonality`	bool	`true`	Include daily seasonality

Optimization Configuration¶

Option	Type	Default	Description
`solver`	string	`"ortools"`	Solver: `ortools`, `pulp`
`stages`	list[str]	`["min_cost_flow"]`	Optimization stages
`constraints_file`	string	`"config/constraints/..."`	Constraints JSON file

Solver Settings¶

Option	Type	Default	Description
`time_limit_seconds`	int	`60`	Max solve time
`optimality_gap`	float	`0.01`	Gap tolerance (1%)
`num_threads`	int	`4`	Parallel threads

Optimization Stages¶

Stage	Description	Phase
`min_cost_flow`	Minimize rebalancing cost	MVP
`critical_demand`	Ensure service levels	Phase 2
`milp_refinement`	Handle discrete constraints	Phase 4

Risk Configuration¶

Option	Type	Default	Description
`model`	string	`"heuristic"`	Model type
`heuristic_threshold`	float	`0.8`	High risk threshold

Heuristic Options¶

Option	Type	Default	Description
`age_weight`	float	`0.3`	Weight for vehicle age
`utilization_weight`	float	`0.4`	Weight for utilization
`maintenance_weight`	float	`0.3`	Weight for maintenance history

Thresholds¶

Option	Type	Default	Description
`high`	float	`0.7`	High risk threshold
`medium`	float	`0.4`	Medium risk threshold
`low`	float	`0.0`	Low risk threshold

API Configuration¶

Option	Type	Default	Description
`host`	string	`"0.0.0.0"`	Bind host
`port`	int	`8000`	Bind port
`log_level`	string	`"INFO"`	Log level
`debug`	bool	`false`	Enable debug mode
`reload`	bool	`false`	Auto-reload on changes

CORS Options¶

Option	Type	Default	Description
`allow_origins`	list[str]	`["*"]`	Allowed origins
`allow_methods`	list[str]	`["*"]`	Allowed methods
`allow_headers`	list[str]	`["*"]`	Allowed headers

Logging Configuration¶

Option	Type	Default	Description
`version`	int	`1`	Config version
`disable_existing_loggers`	bool	`false`	Disable other loggers

Log Levels¶

Level	Value
DEBUG	10
INFO	20
WARNING	30
ERROR	40
CRITICAL	50

Environment Variables¶

Override configuration with environment variables:

# Database
POSTGRES_HOST=localhost
POSTGRES_PORT=5432
POSTGRES_DB=fleet_db
POSTGRES_USER=fleet_user
POSTGRES_PASSWORD=secret

# Cache
REDIS_HOST=localhost
REDIS_PORT=6379

# API
APP_HOST=0.0.0.0
APP_PORT=8000
LOG_LEVEL=INFO

# Data
KAGGLE_USERNAME=username
KAGGLE_KEY=api_key

Next Steps¶

Data Formats - Data schemas
Constraints - Constraint reference