This section provides an overview of the TreeOrdination and the different parameters
of the TreeOrdination class and its methods.
class TreeOrdination.TreeOrdination(feature_names, resampler = NoResample(), metric = "hamming", supervised_clf = ExtraTreesClassifier(1024),
proxy_model = ExtraTreesRegressor(1024), n_iter_unsup = 5, unsup_n_estim = 160, transformer = NoTransform(),
exclude_col = [False, 0], n_neighbors = 8, n_components = 2, min_dist = 0.001, n_jobs = 4)
feature_names: list-like, required
A list of feature names.
resampler: default = NoResample
The re-sampling method to be used. Should follow the format
used by 'imbalaced-learn'. Use this to control how many samples
are passed to LANDMark.
metric: str, default = "hamming"
The metric used by UMAP to calculate the dissimilarity between
LANDMark embeddings.
supervised_clf: default = ExtraTreesClassifier(1024)
The classification model used to predict the class of each sample
using the unsupervised projections.
proxy_model: default = ExtraTreesClassifier(1024)
The regression model used to predict the location of each sample
in the projected space.
landmark_model: default = LANDMarkClassifier(160, use_nnet = False, n_jobs = 8)
n_iter_unsup: int, default = 5
The number of LANDMark embeddings which will be used to construct
the final embedding.
transformer: default = NoTransform()
The pre-processing method used to transform data. Should follow
the 'scikit-learn' format.
exclude_col: list-like, default = [False, [0]]
Specifies which columns should be excluded for scaling and/or
transformation. If the first entry in the list is true the columns
specified by the second entry will be excluded from scaling.
n_neighbors: int, default = 8
The 'n_neighbors' parameter of UMAP. A larger value will capture
more of the global structure of the data while smaller values will
focus more on the local structure of the data. Larger datasets will
likely need a larger value for this parameter.
n_components: int, default = 2
The number of components of the final unsupervised projection.
min_dist: float, default = 0.001
The 'min_dist' parameter of UMAP.
encoded_labels: LabelEncoder
A label encoder used to transform class labels.
X, y: np.ndarray
A saved copy of the training data.
f_name: np.ndarray
The feature names
p_model:
A classifier trained on R_final and the encoded class labels (y)
proj_transformer:
A pre-processing transformer used to transfrom the
the training data, X.
l_model:
A regression model which maps training data to the output produced
by R_PCA (see below).
Rs: list
A list of trained LANDMark models
LM_emb: np.ndarray
A list of LANDMark proximities
transformers: list
A list of pre-processing transformers to be applied prior to fitting
a LANDMark model or using a LANDMark model to calculate proximities.
UMAP_trf: UMAP
A 'UMAP' transformer trained using R_final.
UMAP_emb: np.ndarray
The UMAP transformed data..
PCA_trf: PCA
A 'scikit-learn' PCA transformer fit using the output of the UMAP
transformer.
PCA_emb: np.ndarray
The PCA transformed data.
fit(X, y, **fit_params)
Fits a `TreeOrdination` model.
Parameters:
X: NumPy array of shape (m, n) where 'm' is the number of samples and 'n'
the number of features (taxa, OTUs, ASVs, etc).
y: NumPy array of shape (m,) where 'm' is the number of samples. Each entry
of 'y' should be a factor.
Returns:
A fitted TreeOrdination object.
get_importance(X, y = None)
Prepares an 'alibi' Explainer object.
Parameters:
X: NumPy array of shape (m, n) where 'm' is the number of samples and 'n'
the number of features (taxa, OTUs, ASVs, etc).
y: NumPy array of shape (m,) where 'm' is the number of samples. Each entry
of 'y' should be a factor.
Returns:
A fitted Explainer object.
plot_importance_global(X, y, class_name, n_features = 10, axis = 0, summary = "median", **kwargs)
Produces a plot of per-class global feature importance scores.
Parameters:
X: NumPy array of shape (m, n) where 'm' is the number of samples and 'n'
the number of features (taxa, OTUs, ASVs, etc).
y: NumPy array of shape (m,) where 'm' is the number of samples. Each entry
of 'y' should be a factor.
class_name: str
The name of the class for which importance scores will be plotted.
n_features: int, default = 10
The number of features to plot
axis: int, default = 0
The axis of interest from the R_PCA projection
summary: str, default = "median"
The method used to summarize Shapley values.
Returns:
A 'matplotlib' figure and axis object
plot_importance_local(X, n_features = 10, axis = 0, **kwargs)
Produces a plot of feature importance scores on a per-sample level.
Parameters:
X: NumPy array of shape (m, n) where 'm' is the number of samples and 'n'
the number of features (taxa, OTUs, ASVs, etc).
n_features: int, default = 10
The number of features to plot
axis: int, default = 0
The axis of interest from the R_PCA projection
Returns:
A 'matplotlib' figure and axis object
plot_projection(X, y, ax_1 = 0, ax_2 = 0, use_approx = True, **kwargs)
Produces a plot of feature importance scores on a per-sample level.
Parameters:
X: NumPy array of shape (m, n) where 'm' is the number of samples and 'n'
the number of features (taxa, OTUs, ASVs, etc).
y: NumPy array of shape (m,) where 'm' is the number of samples. Each entry
of 'y' should be a factor.
ax_1: int, default = 0
The first axis of interest from the R_PCA projection
ax_2: int, default = 1
The second axis of interest from the R_PCA projection
use_approx: bool, default = True
Whether the proxy model (l_model from above) or full model will be used
to create the plot for the projection.
Returns:
A 'matplotlib' figure and axis object
predict_proba(X)
Predicts class probabilities.
Input:
X: NumPy array of shape (m, n) where 'm' is the number of samples and 'n'
the number of features (taxa, OTUs, ASVs, etc).
Returns:
A np.ndarray of shape (m, p) where 'm' is the number of samples in X and
'p' is the number of classes.
emb_transform(X, y, trf_type)
Transforms X into a lower dimensional embedding.
Parameters:
X: NumPy array of shape (m, n) where 'm' is the number of samples and 'n'
the number of features (taxa, OTUs, ASVs, etc).
trf_type: str, default = "PCA"
Transforms data using the specified transformer. Options are "approx",
"LM", "UMAP", or "PCA".
Returns:
X_transformed: NumPy array of shape (m, p) where 'm' is the number of samples and 'p'
the number of components specified at the initialization of `TreeOrdination`.
class TreeOrdination.CLRClosureTransform(do_clr = False, delta = None)
do_clr: bool, default = False
Applies the CLR transformation if True. Otherwise only the closure
operation is applied.
fit_transform(X, y = None, **kwargs)
Fits a CLRClosureTransform/NoTransform model and returns the transformed data.
A NoTransform model will just return the original data.
Parameters:
X: NumPy array of shape (m, n) where 'm' is the number of samples and 'n'
the number of features (taxa, OTUs, ASVs, etc).
Returns:
X_transformed: NumPy array of shape (m, n) where 'm' is the number of samples and 'n'
the number of features.
transform(X, y = None, **kwargs)
Takes a CLRClosureTransform/NoTransform model and returns the transformed data.
A NoTransform model will just return the original data.
Parameters:
X: NumPy array of shape (m, n) where 'm' is the number of samples and 'n'
the number of features (taxa, OTUs, ASVs, etc).
Returns:
X_transformed: NumPy array of shape (m, n) where 'm' is the number of samples and 'n'
the number of features.