plspm.plspm module¶

class plspm.plspm.Plspm(data: pandas.core.frame.DataFrame, config: plspm.config.Config, scheme: plspm.scheme.Scheme = <Scheme.CENTROID: <plspm.scheme._CentroidInnerWeightCalculator object>>, iterations: int = 100, tolerance: float = 1e-06, bootstrap: bool = False, bootstrap_iterations: int = 100)¶

Bases: object

Estimates path models with latent variables using partial least squares algorithm

Create an instance of this class in order to estimate a path model using the partial least squares algorithm. When the algorithm has performed the calculations to create the estimate, you can then retrieve the inner and outer models, scores, the path coefficients, effects, and reliability indicators such as goodness-of-fit and unidimensionality. Bootstrapping results can also be retrieved if they were requested.

__init__(data: pandas.core.frame.DataFrame, config: plspm.config.Config, scheme: plspm.scheme.Scheme = <Scheme.CENTROID: <plspm.scheme._CentroidInnerWeightCalculator object>>, iterations: int = 100, tolerance: float = 1e-06, bootstrap: bool = False, bootstrap_iterations: int = 100)¶

Creates an instance of the path model calculator.

Parameters:

data – A Pandas DataFrame containing the dataset to be analyzed
config – An instance of config.Config
scheme – The inner weighting scheme to use: Scheme.CENTROID (default), Scheme.FACTORIAL or Scheme.PATH (see documentation for scheme)
iterations – The maximum number of iterations to try to get the algorithm to converge (default and minimum 100).
tolerance – The tolerance criterion for iterations (default 0.000001, must be >0)
bootstrap – Whether to perform bootstrap validation (default is not to perform validation)
bootstrap_iterations – The number of bootstrap samples to use if bootstrap validation is enabled (default and minimum 100)

Raises:

Exception – if the algorithm cannot converge, or if the requested configuration could not be calculated

bootstrap() → plspm.bootstrap.Bootstrap¶

Gets the results of bootstrap validation, if requested

Returns:	an instance of `bootstrap.Bootstrap` which can be queried for bootstrapping results
Raises:	Exception – if bootstrap validation was not requested or if there were insufficient (<10) observations

crossloadings() → pandas.core.frame.DataFrame¶

Gets the crossloadings

Returns:	a DataFrame with the latent variables as the columns and the manifest variables as the index

effects() → pandas.core.frame.DataFrame¶

Gets direct, indirect, and total effects for each path

Returns:	a DataFrame with an entry in the index for every path in the model, and a column for direct, indirect, and total effects for the corresponding path.

goodness_of_fit() → float¶

Gets goodness-of-fit

Returns:	goodness-of-fit

inner_model() → dict¶

Gets the inner model for the endogenous latent variables

Returns:	a dict with a key for each endogenous latent variable which maps to a DataFrame for its value. The DataFrame for each endogenous latent variable has a row for each latent variable with a path to it, and the columns for estimate, std error, t, and p>\|t\|.

inner_summary() → pandas.core.frame.DataFrame¶

Gets a summary of the inner model

Returns:	a DataFrame with the latent variables as the index, and columns for latent variable type (Exogenous or Endogenous), R squared, block communality, mean redundancy, and AVE (average variance extracted)

outer_model() → pandas.core.frame.DataFrame¶

Gets the outer model

Returns:	a DataFrame with columns for weight, loading, communality, and redundancy, and a row for each manifest variable

path_coefficients() → pandas.core.frame.DataFrame¶

Gets the path coefficient matrix

Returns:	a DataFrame of similar form to the Path matrix passed into `plspm.config.Config`, with the relevant path coefficients in each cell

scores() → pandas.core.frame.DataFrame¶

Gets the latent variable scores

Returns:	a DataFrame with the latent variable scores, with a column for each latent variable. The index is the same as the index of the data passed in.

unidimensionality() → pandas.core.frame.DataFrame¶

Gets the results of checking the unidimensionality of blocks (only meaningful for reflective / mode A blocks)

Returns:	a DataFrame with the latent variables as the index, and columns for Cronbach’s Alpha, Dillon-Goldstein Rho, and the eigenvalues of the first and second principal components.