smart_buildings

Note: This dataset was added recently and is only available in our tfds-nightly package .

• Description:

Smart Buildings Dataset

Dataset accompanying the paper "Real-World Data and Calibrated Simulation Suite for Offline Training of Reinforcement Learning Agents to Optimize Energy and Emission in Office Buildings" by Judah Goldfeder and John Sipple, containing 6 years of detailed telemetric readouts from 3 commercial office buildings.

• Config description: Building sb1

• Homepage: https://github.com/google/sbsim

• Source code: tfds.datasets.smart_buildings.Builder

• Versions:

  • 1.0.0 (default): Initial release.

• Download size: 10.99 GiB

• Dataset size: 86.77 GiB

• Auto-cached (documentation): No

• Splits:

Split	Examples
'sb1_19'	105,120
'sb1_20'	105,408
'sb1_21'	105,120
'sb1_22'	60,480
'sb1_23'	105,120
'sb1_24'	61,344

• Feature structure:

FeaturesDict({
    'action': FeaturesDict({
        'request': FeaturesDict({
            'singleActionRequests': Sequence({
                'continuousValue': Scalar(shape=(), dtype=float32),
                'deviceId': Text(shape=(), dtype=string),
                'setpointName': Text(shape=(), dtype=string),
            }),
            'timestamp': Text(shape=(), dtype=string),
        }),
        'singleActionResponses': Sequence({
            'additionalInfo': Text(shape=(), dtype=string),
            'request': FeaturesDict({
                'continuousValue': Scalar(shape=(), dtype=float32),
                'deviceId': Text(shape=(), dtype=string),
                'setpointName': Text(shape=(), dtype=string),
            }),
            'responseType': Text(shape=(), dtype=string),
        }),
        'timestamp': Text(shape=(), dtype=string),
    }),
    'observation': FeaturesDict({
        'request': FeaturesDict({
            'singleObservationRequests': Sequence({
                'deviceId': Text(shape=(), dtype=string),
                'measurementName': Text(shape=(), dtype=string),
            }),
            'timestamp': Text(shape=(), dtype=string),
        }),
        'singleObservationResponses': Sequence({
            'continuousValue': Scalar(shape=(), dtype=float32),
            'observationValid': Text(shape=(), dtype=string),
            'singleObservationRequest': FeaturesDict({
                'deviceId': Text(shape=(), dtype=string),
                'measurementName': Text(shape=(), dtype=string),
            }),
            'timestamp': Text(shape=(), dtype=string),
        }),
        'timestamp': Text(shape=(), dtype=string),
    }),
    'reward': FeaturesDict({
        'agentRewardValue': Scalar(shape=(), dtype=float32),
        'carbonEmissionWeight': Scalar(shape=(), dtype=float32),
        'carbonEmitted': Scalar(shape=(), dtype=float32),
        'electricityEnergyCost': Scalar(shape=(), dtype=float32),
        'endTimestamp': Text(shape=(), dtype=string),
        'energyCostWeight': Scalar(shape=(), dtype=float32),
        'naturalGasEnergyCost': Scalar(shape=(), dtype=float32),
        'normalizedCarbonEmission': Scalar(shape=(), dtype=float32),
        'normalizedEnergyCost': Scalar(shape=(), dtype=float32),
        'normalizedProductivityRegret': Scalar(shape=(), dtype=float32),
        'personProductivity': Scalar(shape=(), dtype=float32),
        'productivityRegret': Scalar(shape=(), dtype=float32),
        'productivityReward': Scalar(shape=(), dtype=float32),
        'productivityWeight': Scalar(shape=(), dtype=float32),
        'rewardScale': Scalar(shape=(), dtype=float32),
        'startTimestamp': Text(shape=(), dtype=string),
        'totalOccupancy': Scalar(shape=(), dtype=float32),
    }),
})

• Feature documentation:

Feature	Class	Shape	Dtype	Description
	FeaturesDict
action	FeaturesDict
action/request	FeaturesDict
action/request/singleActionRequests	Sequence
action/request/singleActionRequests/continuousValue	Scalar		float32
action/request/singleActionRequests/deviceId	Text		string
action/request/singleActionRequests/setpointName	Text		string
action/request/timestamp	Text		string
action/singleActionResponses	Sequence
action/singleActionResponses/additionalInfo	Text		string
action/singleActionResponses/request	FeaturesDict
action/singleActionResponses/request/continuousValue	Scalar		float32
action/singleActionResponses/request/deviceId	Text		string
action/singleActionResponses/request/setpointName	Text		string
action/singleActionResponses/responseType	Text		string
action/timestamp	Text		string
observation	FeaturesDict
observation/request	FeaturesDict
observation/request/singleObservationRequests	Sequence
observation/request/singleObservationRequests/deviceId	Text		string
observation/request/singleObservationRequests/measurementName	Text		string
observation/request/timestamp	Text		string
observation/singleObservationResponses	Sequence
observation/singleObservationResponses/continuousValue	Scalar		float32
observation/singleObservationResponses/observationValid	Text		string
observation/singleObservationResponses/singleObservationRequest	FeaturesDict
observation/singleObservationResponses/singleObservationRequest/deviceId	Text		string
observation/singleObservationResponses/singleObservationRequest/measurementName	Text		string
observation/singleObservationResponses/timestamp	Text		string
observation/timestamp	Text		string
reward	FeaturesDict
reward/agentRewardValue	Scalar		float32
reward/carbonEmissionWeight	Scalar		float32
reward/carbonEmitted	Scalar		float32
reward/electricityEnergyCost	Scalar		float32
reward/endTimestamp	Text		string
reward/energyCostWeight	Scalar		float32
reward/naturalGasEnergyCost	Scalar		float32
reward/normalizedCarbonEmission	Scalar		float32
reward/normalizedEnergyCost	Scalar		float32
reward/normalizedProductivityRegret	Scalar		float32
reward/personProductivity	Scalar		float32
reward/productivityRegret	Scalar		float32
reward/productivityReward	Scalar		float32
reward/productivityWeight	Scalar		float32
reward/rewardScale	Scalar		float32
reward/startTimestamp	Text		string
reward/totalOccupancy	Scalar		float32

• Supervised keys (See as_supervised doc): None

• Figure (tfds.show_examples): Not supported.

• Examples (tfds.as_dataframe):

• Citation:

// TODO(smart_buildings_dataset): BibTeX citation

smart_buildings/sb1 (default config)