Data Dictionary

This comprehensive reference describes all data fields available in Enertel forecast extracts and API responses. During your trial, you'll receive historical backcast data for your price nodes of interest that follows this schema.

Data Structure Overview

Our forecast data is structured around the core concepts described in our concepts documentation. Each row in your data extract represents a single forecast point with associated metadata and percentile values.

Field Reference

Object Information

Column	Type	Description
object_name	string	Name of the forecasted object, typically a price node from the ISO
object_type	string	Classification of the object (e.g., 'price_node', 'hub', 'interface')
object_subtype	string	Additional classification details for the object
object_latitude	number	Latitude coordinate of the object location
object_longitude	number	Longitude coordinate of the object location
object_iso	string	ISO code representing the geographic market area

Power Plant Information

Available when forecasting generation assets

Column	Type	Description
power_plant	string	Name of the power plant (if applicable)
plant_technology	string	Generation technology (e.g., 'natural_gas', 'solar', 'wind')
prime_mover_code	string	Primary equipment type code (e.g., turbine type)
plant_capacity	number	Nameplate capacity of the power plant in megawatts

Target and Series Information

Column	Type	Description
target_id	integer	Unique identifier for the forecast target
description	string	Human-readable description of the target (e.g., "ERCOT DALMP")
timezone	string	Timezone for all timestamps in this target
iso	string	Independent System Operator code
series_name	string	Name of the data series being forecasted

Scenario and Batch Information

Column	Type	Description
scenario_id	integer	Unique identifier for the input scenario
range_start	string	Start of forecast time range (ISO 8601 format)
range_end	string	End of forecast time range (ISO 8601 format)
schedule_id	integer	Identifier for the processing schedule
scheduled_at	string	When the forecast batch was scheduled (ISO 8601 format)
batch_id	integer	Unique identifier for the forecast batch
batch_created_at	string	When the forecast batch was created (ISO 8601 format)

Model Information

Column	Type	Description
model_id	integer	Unique identifier for the forecasting model
model_created_at	string	When the model was trained/created (ISO 8601 format)

Forecast Data

Column	Type	Description
timestamp	string	The exact time for which the forecast applies (ISO 8601 format)
value	number	Expected value (mean) of the forecast distribution
feature_id	integer	Unique identifier for the specific feature being forecasted

Probabilistic Forecasts

Our models provide complete probability distributions through percentile forecasts:

Column	Type	Description
p01	number	1st percentile - extreme low outcomes (1% chance of being lower)
p05	number	5th percentile - very low outcomes
p10	number	10th percentile - low outcomes
p20	number	20th percentile
p30	number	30th percentile
p40	number	40th percentile
p50	number	Median forecast - 50th percentile (most commonly used)
p60	number	60th percentile
p70	number	70th percentile
p80	number	80th percentile
p90	number	90th percentile - high outcomes
p95	number	95th percentile - very high outcomes
p99	number	99th percentile - extreme high outcomes (1% chance of being higher)

Validation Data

Available in historical backcasts

Column	Type	Description
actual	number	The actual observed value for validation and accuracy assessment
error	number	Difference between forecasted value and actual value (forecast - actual)

Data Quality Notes

Timestamp Handling

• All timestamps are provided in ISO 8601 format
• Timestamps are automatically converted to the target's local timezone
• Be aware of daylight saving time transitions when working with time series data

Missing Data

• Our data extracts are cleaned and validated before delivery
• Any missing values are explicitly handled and documented
• Contact support if you encounter unexpected null values

Duplicate Detection

The composite primary key for forecast data is:

(timestamp, feature_id, scenario_id, model_id)

This combination should be unique across your dataset.

Common Usage Patterns

Basic Forecast Analysis

# Most common columns for basic analysis
essential_cols = [
    'object_name', 'series_name', 'timestamp', 
    'scheduled_at', 'p50', 'actual'
]

Probabilistic Analysis

# All percentile columns for uncertainty analysis
percentile_cols = [f'p{i:02d}' for i in [1, 5, 10, 20, 30, 40, 50, 60, 70, 80, 90, 95, 99]]

Model Comparison

# Key columns for comparing different models
comparison_cols = [
    'model_id', 'model_created_at', 'feature_id', 
    'timestamp', 'p50', 'actual'
]

Units and Conventions

• Prices: Typically in $/MWh
• Generation: Typically in MW or MWh depending on the series
• Timestamps: Always in the target's local timezone
• Coordinates: Decimal degrees (WGS84)

For specific units in your data extract, refer to the series_name and description fields which provide context about the measurement type and units.