Methods

Metalearners

S-Learner

class reina.metalearners.SLearner

Methods

Function Name Description
init Initialize object.
effects Calculates and returns the treatment effects from this class.
fit Train the causal model.

effects(self, X, treatment)

Calculates and returns the treatment effects from this class.

Parameters

  • X (2D Spark Dataframe): Feature data to estimate treatment effect of
  • treatment (Str): Name of the treatment variable

Returns

  • cate (2D Spark DataFrame): conditional average treatment effect
  • ate (float): average treatment effect

fit(self, data, treatments, outcome, estimator)

Trains the ML-based causal model for this class.

Parameters

  • data (2-D Spark dataframe): Base dataset containing features, treatment, iv, and outcome variables
  • treatments (List): Names of the treatment variables
  • outcome (Str): Name of the outcome variable
  • estimator (sklearn model obj): Arbitrary ML estimator of choice

Returns

  • None (self)

Example

Example S-learner usage can be found here.

T-Learner

class reina.metalearners.TLearner

Methods

Function Name Description
init Initialize object.
effects Calculates and returns the treatment effects from this class.
fit Trains the causal model

effects(self, X, treatment)

Calculates and returns the treatment effects from this class.

Parameters

  • X (2D Spark Dataframe): Feature data to estimate treatment effect of
  • treatment (Str): Name of the treatment variable

Returns

  • cate (2D Spark DataFrame): conditional average treatment effect
  • ate (float): average treatment effect

fit(self, data, treatments, outcome, estimator_0, estimator_1)

Trains the ML-based causal model for this class.

Parameters

  • data (2-D Spark dataframe): Base dataset containing features, treatment, iv, and outcome variables
  • treatments (List): Names of the treatment variables
  • outcome (Str): Name of the outcome variable
  • estimator_0 (mllib model obj): Arbitrary ML model of choice
  • estimator_1 (mllib model obj): Arbitrary ML model of choice

Returns

  • None (self)

Example

Example T-learner usage can be found here.

X-Learner

class reina.metalearners.XLearner

Methods

Function Name Description
init Initialize object.
effects Calculates and returns the treatment effects from this class.
fit Trains the causal model.

effects(self, X, treatment)

Calculates and returns the treatment effects from this class.

Parameters

  • X (2D Spark Dataframe): Feature data to estimate treatment effect of
  • treatment (Str): Name of the treatment variable

Returns

  • cate (2D Spark DataFrame): conditional average treatment effect
  • ate (float): average treatment effect

fit(self, data, treatments, outcome, estimator_0, estimator_1)

Trains the ML-based causal model for this class.

Parameters

  • data (2-D Spark dataframe): Base dataset containing features, treatment, iv, and outcome variables
  • treatments (List): Names of the treatment variables
  • outcome (Str): Name of the outcome variable
  • estimator_10 (mllib model obj): Arbitrary ML model of choice
  • estimator_11 (mllib model obj): Arbitrary ML model of choice
  • estimator_20 (mllib model obj): Arbitrary ML model of choice
  • estimator_21 (mllib model obj): Arbitrary ML model of choice
  • propensity_estimator (mllib model obj): Arbitrary ML model for propensity function

Returns

  • None (self)

Example

Example X-learner usage can be found here.

IV-based Methods

Two-Stage Least Squares

class reina.iv.TwoStageLeastSquares

Methods

Function Name Description
init Initialize object.
effects Calculates and returns the treatment effects from this class.
fit Trains the causal model.

effects(self, X, treatment)

Calculates and returns the treatment effects from this class.

Parameters

  • X (2D Spark Dataframe): Feature data to estimate treatment effect of
  • treatment (Str): Name of the treatment variable

Returns

  • cate (2D Spark DataFrame): conditional average treatment effect
  • ate (float): average treatment effect

fit(self, data, data, treatments, outcome, iv)

Trains the ML-based causal model for this class.

Parameters

  • data (2-D Spark dataframe): Base dataset containing features, treatment, iv, and outcome variables
  • treatments (List): Names of the treatment variables
  • outcome (Str): Name of the outcome variable
  • iv (Str): Name of the instrument variable

Returns

  • None (self)

Example

Example TwoStageLeastSquares usage can be found here.