neighbayes.models.Level

class neighbayes.models.Level(formula=None, data=None, *, y=None, X=None, W=None, process='SAR', key=None, groups=None, name=None)[source]

One level’s equation for Multilevel.

Parameters:
formula : str, optional

At the units, "y ~ x1 + x2". Above them, the level’s covariates, "~ z1 + z2"; the intercept is dropped (the units carry the only one). None or "~ 1" means no covariates.

data : DataFrame, optional

The level’s rows. At the units, the data the formula reads. Above, one row per group, matched to the groups by key (or the index).

y : array-like, optional

Matrix mode in place of formula/data (y at the units only).

X : array-like, optional

Matrix mode in place of formula/data (y at the units only).

W : libpysal.graph.Graph or scipy.sparse matrix

The level’s graph over all its groups, regardless of parent. Optional when process=None.

process : {"SAR", "SEM", "SDM", "SDEM", "SLX", None}, default "SAR"

The level’s spatial process; the Durbin processes add the spatial lags of the level’s covariates, W_ℓ X_ℓ. None is aspatial.

key : str, optional

The id column shared with the level below: each lower row’s value names its group here. Looked up in the level below’s data, then in the units’ data (strict nesting is checked).

groups : array-like, optional

Matrix mode in place of key: one group label per row of the level below.

name : str, optional

A label for summaries; defaults to key.

Notes

The groups are ordered by the Graph’s ids when W is a Graph (the level’s data are reindexed to them); otherwise by the rows of data, or by the sorted labels when there are no data. A sparse W must follow that order.

__init__(formula=None, data=None, *, y=None, X=None, W=None, process='SAR', key=None, groups=None, name=None)[source]

Methods

__init__([formula, data, y, X, W, process, ...])

Attributes

W

X

data

formula

groups

key

name

process

y

W = None[source]
X = None[source]
data = None[source]
formula = None[source]
groups = None[source]
key = None[source]
name = None[source]
process = 'SAR'[source]
y = None[source]