def predict_proba(self, X):
"""Predicts the target probability of an entire dataset contained in memory.
Parameters
----------
X
array-like of shape (n_samples, n_features).
Returns
-------
Predicted target values for each row of `X`.
"""
# Check the fit method has been called
utils.validation.check_is_fitted(self, attributes='instance_')
# Check the input
X = utils.check_array(X, **SKLEARN_INPUT_X_PARAMS)
if X.shape[1] != self.n_features_in_:
raise ValueError(
f'Expected {self.n_features_in_} features, got {X.shape[1]}')
# river's predictions have to converted to follow the scikit-learn conventions
def reshape_probas(y_pred):
return [y_pred.get(c, 0) for c in self.classes_]
# Make a prediction for each observation
y_pred = np.empty(shape=(len(X), len(self.classes_)))
for i, (x, _) in enumerate(stream.iter_array(X)):
y_pred[i] = reshape_probas(self.instance_.predict_proba_one(x))
return y_pred
def predict(self, X):
"""Predicts the target of an entire dataset contained in memory.
Parameters
----------
X
array-like of shape (n_samples, n_features).
Returns
-------
Predicted target values for each row of `X`.
"""
# Check the fit method has been called
utils.validation.check_is_fitted(self, attributes='instance_')
# Check the input
X = utils.check_array(X, **SKLEARN_INPUT_X_PARAMS)
if X.shape[1] != self.n_features_in_:
raise ValueError(
f'Expected {self.n_features_in_} features, got {X.shape[1]}')
# Make a prediction for each observation
y_pred = [None] * len(X)
for i, (x, _) in enumerate(stream.iter_array(X)):
y_pred[i] = self.instance_.predict_one(x)
# Convert back to the expected labels if an encoder was necessary for binary classification
y_pred = np.asarray(y_pred)
if hasattr(self, 'label_encoder_'):
y_pred = self.label_encoder_.inverse_transform(y_pred.astype(int))
return y_pred
def predict(self, X) -> np.ndarray:
"""Predicts the target of an entire dataset contained in memory.
Parameters
----------
X
array-like of shape (n_samples, n_features).
Returns
-------
Predicted target values for each row of `X`.
"""
# Check the fit method has been called
utils.validation.check_is_fitted(self, attributes='instance_')
# Check the input
X = utils.check_array(X, **SKLEARN_INPUT_X_PARAMS)
if X.shape[1] != self.n_features_in_:
raise ValueError(
f'Expected {self.n_features_in_} features, got {X.shape[1]}')
# Make a prediction for each observation
y_pred = np.empty(shape=len(X))
for i, (x, _) in enumerate(stream.iter_array(X)):
y_pred[i] = self.instance_.predict_one(x)
return y_pred
def iter_sklearn_dataset(dataset: "sklearn.utils.Bunch",
**kwargs) -> base.typing.Stream:
"""Iterates rows from one of the datasets provided by scikit-learn.
This allows you to use any dataset from [scikit-learn's `datasets` module](https://scikit-learn.org/stable/modules/classes.html#module-sklearn.datasets). For instance, you can use the `fetch_openml` function to get access to all of the
datasets from the OpenML website.
Parameters
----------
dataset
A scikit-learn dataset.
kwargs
Extra keyword arguments are passed to the underlying call to `stream.iter_array`.
Examples
--------
>>> import pprint
>>> from sklearn import datasets
>>> from river import stream
>>> dataset = datasets.load_boston()
>>> for xi, yi in stream.iter_sklearn_dataset(dataset):
... pprint.pprint(xi)
... print(yi)
... break
{'AGE': 65.2,
'B': 396.9,
'CHAS': 0.0,
'CRIM': 0.00632,
'DIS': 4.09,
'INDUS': 2.31,
'LSTAT': 4.98,
'NOX': 0.538,
'PTRATIO': 15.3,
'RAD': 1.0,
'RM': 6.575,
'TAX': 296.0,
'ZN': 18.0}
24.0
"""
kwargs["X"] = dataset.data
kwargs["y"] = dataset.target
try:
kwargs["feature_names"] = dataset.feature_names
except AttributeError:
pass
if isinstance(kwargs["X"], pd.DataFrame):
yield from stream.iter_pandas(**kwargs)
else:
yield from stream.iter_array(**kwargs)
def iter_sklearn_dataset(dataset: "sklearn.utils.Bunch",
**kwargs) -> base.typing.Stream:
"""Iterates rows from one of the datasets provided by scikit-learn.
This allows you to use any dataset from [scikit-learn's `datasets` module](https://scikit-learn.org/stable/modules/classes.html#module-sklearn.datasets). For instance, you can use the `fetch_openml` function to get access to all of the
datasets from the OpenML website.
Parameters
----------
dataset
A scikit-learn dataset.
kwargs
Extra keyword arguments are passed to the underlying call to `stream.iter_array`.
Examples
--------
>>> import pprint
>>> from sklearn import datasets
>>> from river import stream
>>> dataset = datasets.load_diabetes()
>>> for xi, yi in stream.iter_sklearn_dataset(dataset):
... pprint.pprint(xi)
... print(yi)
... break
{'age': 0.0380759064334241,
'bmi': 0.0616962065186885,
'bp': 0.0218723549949558,
's1': -0.0442234984244464,
's2': -0.0348207628376986,
's3': -0.0434008456520269,
's4': -0.00259226199818282,
's5': 0.0199084208763183,
's6': -0.0176461251598052,
'sex': 0.0506801187398187}
151.0
"""
kwargs["X"] = dataset.data
kwargs["y"] = dataset.target
try:
kwargs["feature_names"] = dataset.feature_names
except AttributeError:
pass
if isinstance(kwargs["X"], pd.DataFrame):
yield from stream.iter_pandas(**kwargs)
else:
yield from stream.iter_array(**kwargs)
def iter_pandas(X: pd.DataFrame,
y: typing.Union[pd.Series, pd.DataFrame] = None,
**kwargs) -> base.typing.Stream:
"""Iterates over the rows of a `pandas.DataFrame`.
Parameters
----------
X
A dataframe of features.
y
A series or a dataframe with one column per target.
kwargs
Extra keyword arguments are passed to the underlying call to `stream.iter_array`.
Examples
--------
>>> import pandas as pd
>>> from river import stream
>>> X = pd.DataFrame({
... 'x1': [1, 2, 3, 4],
... 'x2': ['blue', 'yellow', 'yellow', 'blue'],
... 'y': [True, False, False, True]
... })
>>> y = X.pop('y')
>>> for xi, yi in stream.iter_pandas(X, y):
... print(xi, yi)
{'x1': 1, 'x2': 'blue'} True
{'x1': 2, 'x2': 'yellow'} False
{'x1': 3, 'x2': 'yellow'} False
{'x1': 4, 'x2': 'blue'} True
"""
kwargs['feature_names'] = X.columns
if isinstance(y, pd.DataFrame):
kwargs['target_names'] = y.columns
yield from stream.iter_array(X=X.to_numpy(),
y=y if y is None else y.to_numpy(),
**kwargs)
from river import compose
from river import linear_model
from river import metrics
from river import evaluate
from river import preprocessing
from river import optim
from river import stream
import river
model = preprocessing.StandardScaler()
model |= linear_model.LinearRegression(optimizer=optim.SGD(0.001))
s2 = stream.iter_array(Xa)
audit = [model.predict_one(i[0]) for i in s2]
plt.imshow(reshape_vals(audit))
s1 = stream.iter_array(Xc, a0.ev(contexts))
for c, v in s1:
model.learn_one(c, v)
s2 = stream.iter_array(Xa)
audit = [model.predict_one(i[0]) for i in s2]
plt.imshow(reshape_vals(audit))
metric = metrics.RMSE()
evaluate.progressive_val_score(stream.iter_array(Xc, a0.ev(contexts)), model, metric, print_every=int(Xc.shape[0]/20))
s2 = stream.iter_array(Xa)
