[REVIEW][PROPOSAL] Add tags and prefered memory order tags to estimators

python/cuml/linear_model/elastic_net.pyx

@@ -134,12 +134,17 @@ class ElasticNet(Base, RegressorMixin):
         The estimated coefficients for the linear regression model.
     intercept_ : array
         The independent term. If `fit_intercept` is False, will be 0.
+    preferred_order : 'F'
+        Whether the algorithm internally operates in column ('F') or row ('C')
+        major data. Pass data to the algorithm with this order for the best
+        memory and time processing efficiency.
 
     Notes
     -----
     For additional docs, see `scikitlearn's ElasticNet
     <https://scikit-learn.org/stable/modules/generated/sklearn.linear_model.ElasticNet.html>`_.
     """
+    preferred_order = 'F'
 
     def __init__(self, alpha=1.0, l1_ratio=0.5, fit_intercept=True,
                  normalize=False, max_iter=1000, tol=1e-3, selection='cyclic',

python/cuml/linear_model/lasso.pyx

@@ -130,12 +130,17 @@ class Lasso(Base, RegressorMixin):
         The estimated coefficients for the linear regression model.
     intercept_ : array
         The independent term. If `fit_intercept` is False, will be 0.
+    preferred_order : 'F'
+        Whether the algorithm internally operates in column ('F') or row ('C')
+        major data. Pass data to the algorithm with this order for the best
+        memory and time processing efficiency.
 
     Notes
     -----
     For additional docs, see `scikitlearn's Lasso
     <https://scikit-learn.org/stable/modules/generated/sklearn.linear_model.Lasso.html>`_.
     """
+    preferred_order = 'F'
 
     def __init__(self, alpha=1.0, fit_intercept=True, normalize=False,
                  max_iter=1000, tol=1e-3, selection='cyclic', handle=None,

python/cuml/linear_model/linear_regression.pyx

@@ -171,6 +171,10 @@ class LinearRegression(Base, RegressorMixin):
         The estimated coefficients for the linear regression model.
     intercept_ : array
         The independent term. If `fit_intercept` is False, will be 0.
+    preferred_order : 'F'
+        Whether the algorithm internally operates in column ('F') or row ('C')
+        major data. Pass data to the algorithm with this order for the best
+        memory and time processing efficiency.
 
     Notes
     ------
@@ -196,6 +200,7 @@ class LinearRegression(Base, RegressorMixin):
 
 
     """
+    preferred_order = 'F'
 
     def __init__(self, algorithm='eig', fit_intercept=True, normalize=False,
                  handle=None, verbose=False, output_type=None):

python/cuml/linear_model/logistic_regression.pyx

@@ -156,6 +156,10 @@ class LogisticRegression(Base, ClassifierMixin):
         is True
     intercept_: device array (n_classes, 1)
         The independent term. If `fit_intercept` is False, will be 0.
+    preferred_order : 'F'
+        Whether the algorithm internally operates in column ('F') or row ('C')
+        major data. Pass data to the algorithm with this order for the best
+        memory and time processing efficiency.
 
     Notes
     ------
@@ -169,6 +173,7 @@ class LogisticRegression(Base, ClassifierMixin):
     For additional information, see Scikit-learn's LogistRegression
     <https://scikit-learn.org/stable/modules/generated/sklearn.linear_model.LogisticRegression.html>`_.
     """
+    preferred_order = 'F'
 
     def __init__(
         self,

python/cuml/linear_model/mbsgd_classifier.pyx

@@ -151,6 +151,8 @@ class MBSGDClassifier(Base, ClassifierMixin):
     <https://scikit-learn.org/stable/modules/generated/sklearn.linear_model.SGDClassifier.html>`_.
     """
 
+    preferred_order = 'F'
+
     def __init__(self, loss='hinge', penalty='l2', alpha=0.0001,
                  l1_ratio=0.15, fit_intercept=True, epochs=1000, tol=1e-3,
                  shuffle=True, learning_rate='constant', eta0=0.001,

python/cuml/linear_model/mbsgd_regressor.pyx

@@ -142,6 +142,8 @@ class MBSGDRegressor(Base, RegressorMixin):
     <https://scikit-learn.org/stable/modules/generated/sklearn.linear_model.SGDRegressor.html>`_.
     """
 
+    preferred_order = 'F'
+
     def __init__(self, loss='squared_loss', penalty='l2', alpha=0.0001,
                  l1_ratio=0.15, fit_intercept=True, epochs=1000, tol=1e-3,
                  shuffle=True, learning_rate='constant', eta0=0.001,

python/cuml/linear_model/ridge.pyx

@@ -188,6 +188,10 @@ class Ridge(Base, RegressorMixin):
         The estimated coefficients for the linear regression model.
     intercept_ : array
         The independent term. If `fit_intercept` is False, will be 0.
+    preferred_order : 'F'
+        Whether the algorithm internally operates in column ('F') or row ('C')
+        major data. Pass data to the algorithm with this order for the best
+        memory and time processing efficiency.
 
     Notes
     ------
@@ -207,6 +211,8 @@ class Ridge(Base, RegressorMixin):
     <https://scikit-learn.org/stable/modules/generated/sklearn.linear_model.Ridge.html>`_.
     """
 
+    preferred_order = 'F'
+
     def __init__(self, alpha=1.0, solver='eig', fit_intercept=True,
                  normalize=False, handle=None, output_type=None,
                  verbose=False):

python/cuml/solvers/cd.pyx

@@ -186,6 +186,7 @@ class CD(Base):
         See :ref:`output-data-type-configuration` for more info.
 
     """
+    preferred_order = 'F'
 
     def __init__(self, loss='squared_loss', alpha=0.0001, l1_ratio=0.15,
                  fit_intercept=True, normalize=False, max_iter=1000, tol=1e-3,

python/cuml/solvers/qn.pyx

@@ -232,6 +232,10 @@ class QN(Base):
         Note: shape is (n_classes, n_features + 1) if fit_intercept = True.
     intercept_ : array (n_classes, 1)
         The independent term. If `fit_intercept` is False, will be 0.
+    preferred_order : 'F'
+        Whether the algorithm internally operates in column ('F') or row ('C')
+        major data. Pass data to the algorithm with this order for the best
+        memory and time processing efficiency.
 
     Notes
     ------
@@ -244,6 +248,7 @@ class QN(Base):
          2007]
          <https://www.microsoft.com/en-us/research/publication/scalable-training-of-l1-regularized-log-linear-models/>
     """
+    preferred_order = 'F'
 
     def __init__(self, loss='sigmoid', fit_intercept=True,
                  l1_strength=0.0, l2_strength=0.0, max_iter=1000, tol=1e-3,

python/cuml/solvers/sgd.pyx

@@ -217,6 +217,8 @@ class SGD(Base):
 
     """
 
+    preferred_order = 'F'
+
     def __init__(self, loss='squared_loss', penalty='none', alpha=0.0001,
                  l1_ratio=0.15, fit_intercept=True, epochs=1000, tol=1e-3,
                  shuffle=True, learning_rate='constant', eta0=0.001,