Bayesian


The bayesian binary sensor platform observes the state from multiple sensors and uses Bayes’ rule to estimate the probability that an event has occurred given the state of the observed sensors. If the estimated posterior probability is above the probability_threshold, the sensor is on otherwise it is off.

This allows for the detection of complex events that may not be readily observable, e.g., cooking, showering, in bed, the start of a morning routine, etc. It can also be used to gain greater confidence about events that are directly observable, but for which the sensors can be unreliable, e.g., presence.

Configuration

To enable the Bayesian sensor, add the following lines to your configuration.yaml:

# Example configuration.yaml entry
binary_sensor:
  - platform: bayesian
    prior: 0.1
    observations:
      - entity_id: 'switch.kitchen_lights'
        prob_given_true: 0.6
        prob_given_false: 0.2
        platform: 'state'
        to_state: 'on'

Configuration Variables

prior float Required

The prior probability of the event. At any point in time (ignoring all external influences) how likely is this event to occur?

probability_threshold float (Optional, default: 0.5)

The probability at which the sensor should trigger to on.

name string (Optional, default: Bayesian Binary Sensor)

Name of the sensor to use in the frontend.

observations list Required

The observations which should influence the likelihood that the given event has occurred.

platform string Required

The supported platforms are state, numeric_state, and template. They are modeled after their corresponding triggers for automations, requiring to_state (for state), below and/or above (for numeric_state) and value_template (for template).

entity_id string (Optional)

Name of the entity to monitor. Required for state and numeric_state.

value_template template (Optional)

Defines the template to be used. Required for template.

prob_given_true float Required

The probability of the observation occurring, given the event is true.

prob_given_false float (Optional)

The probability of the observation occurring, given the event is false can be set as well.

Default:

1 - prob_given_true if prob_given_false is not set

to_state string (Optional)

The target state. Required (for state).

Full examples

The following is an example for the state observation platform.

# Example configuration.yaml entry
binary_sensor:
  name: 'in_bed'
  platform: 'bayesian'
  prior: 0.25
  probability_threshold: 0.95
  observations:
    - platform: 'state'
      entity_id: 'sensor.living_room_motion'
      prob_given_true: 0.4
      prob_given_false: 0.2
      to_state: 'off'
    - platform: 'state'
      entity_id: 'sensor.basement_motion'
      prob_given_true: 0.5
      prob_given_false: 0.4
      to_state: 'off'
    - platform: 'state'
      entity_id: 'sensor.bedroom_motion'
      prob_given_true: 0.5
      to_state: 'on'
    - platform: 'state'
      entity_id: 'sun.sun'
      prob_given_true: 0.7
      to_state: 'below_horizon'

Next up an example which targets the numeric_state observation platform, as seen in the configuration it requires below and/or above instead of to_state.

# Example configuration.yaml entry
binary_sensor:
  name: 'Heat On'
  platform: 'bayesian'
  prior: 0.2
  probability_threshold: 0.9
  observations:
    - platform: 'numeric_state'
      entity_id: 'sensor.outside_air_temperature_fahrenheit'
      prob_given_true: 0.95
      below: 50

Finally, here’s an example for template observation platform, as seen in the configuration it requires value_template.

# Example configuration.yaml entry
binary_sensor:
  name: 'Paulus Home'
  platform: 'bayesian'
  prior: 0.5
  probability_threshold: 0.9
  observations:
    - platform: template
      value_template: >
        {{is_state('device_tracker.paulus','not_home') and ((as_timestamp(now()) - as_timestamp(states.device_tracker.paulus.last_changed)) > 300)}}
      prob_given_true: 0.95

The template is re-evaluated whenever an entity ID that it references changes state. If you use non-deterministic functions like now() in the template it will not be continuously re-evaluated, but only when an entity ID that is referenced is updated.

In this example, since the template is only evaluated on state change of device_tracker.paulus the template won’t change state after 5 mins like intended. The ways to force template reevaluation are documented in the template binary_sensor.