First line of docstrings for models and optimizers (QB3#120)

Author: mathurinm

doc/api.rst

@@ -58,11 +58,11 @@ Criterion
 Algorithms
 ==========
 
-:py:mod:`sparse_ho`:
+:py:mod:`sparse_ho.algo`:
 
-.. currentmodule:: sparse_ho
+.. currentmodule:: sparse_ho.algo
 
-.. automodule:: sparse_ho
+.. automodule:: sparse_ho.algo
    :no-members:
    :no-inherited-members:
 
@@ -78,11 +78,11 @@ Algorithms
 Optimizers
 ==========
 
-:py:mod:`sparse_ho`:
+:py:mod:`sparse_ho.optimizers`:
 
-.. currentmodule:: sparse_ho
+.. currentmodule:: sparse_ho.optimizers
 
-.. automodule:: sparse_ho
+.. automodule:: sparse_ho.optimizers
    :no-members:
    :no-inherited-members:
 

sparse_ho/models/enet.py

@@ -11,7 +11,7 @@
 
 
 class ElasticNet(BaseModel):
-    """Sparse ho ElasticNet model (inner problem)
+    """Sparse ho ElasticNet model (inner problem).
 
     Parameters
     ----------

sparse_ho/models/lasso.py

@@ -11,7 +11,8 @@
 
 
 class Lasso(BaseModel):
-    """Linear Model trained with L1 prior as regularizer (aka the Lasso)
+    """Linear Model trained with L1 prior as regularizer (aka the Lasso).
+
     The optimization objective for Lasso is:
     (1 / (2 * n_samples)) * ||y - Xw||^2_2 + alpha * ||w||_1
 

sparse_ho/models/logreg.py

@@ -9,6 +9,7 @@
 
 class SparseLogreg(BaseModel):
     """Sparse Logistic Regression classifier.
+
     The objective function is:
 
     sum_1^n_samples log(1 + e^{-y_i x_i^T w}) + 1. / C * ||w||_1

sparse_ho/models/svm.py

@@ -12,10 +12,11 @@
 
 
 class SVM(BaseModel):
-    """The support vector Machine classifier without bias
+    """Support Vector Machine classifier without bias.
+
     The optimization problem is solved in the dual:
-    1/2 r^T(y * X)(y * X)^T r - sum_i^n r_i
-    s.t 0 <= r_i <= C
+        1/2 r^T(y * X)(y * X)^T r - sum_i^n r_i
+        s.t 0 <= r_i <= C
 
     Parameters
     ----------

sparse_ho/models/svr.py

@@ -81,7 +81,8 @@ def _update_beta_jac_bcd_aux_sparse(data, indptr, indices, y, epsilon, beta,
 
 
 class SVR(BaseModel):
-    """The support vector regression without bias
+    """The support vector regression without bias.
+
     The optimization problem is solved in the dual.
 
     Parameters

sparse_ho/models/wlasso.py

@@ -10,7 +10,7 @@
 
 
 class WeightedLasso(BaseModel):
-    r"""Linear Model trained with weighted L1 regularizer (aka weighted Lasso)
+    r"""Linear Model trained with weighted L1 regularizer (aka weighted Lasso).
 
     The optimization objective for weighted Lasso is:
 

sparse_ho/optimizers/adam.py

@@ -5,16 +5,18 @@
 
 
 class Adam(BaseOptimizer):
-    """This Adam code is taken from
+    """ADAM optimizer for the outer problem.
+
+    This Adam code is taken from
     https://github.com/sagarvegad/Adam-optimizer/blob/master/Adam.py
 
     Parameters
     ----------
     n_outer: int, optional (default=100).
-        number of maximum updates of alpha.
+        Number of maximum updates of alpha.
     epsilon: float, optional (default=1e-3)
     lr: float, optional (default=1e-2)
-        learning rate
+        Learning rate
     beta_1: float, optional (default=0.9)
     beta_2: float, optional (default=0.999)
     verbose: bool, optional (default=False)

sparse_ho/optimizers/line_search.py

@@ -5,7 +5,9 @@
 
 
 class LineSearch(BaseOptimizer):
-    """This line-search code is taken from here:
+    """Gradient descent with line search for the outer problem.
+
+    The code is taken from here:
     https://github.com/fabianp/hoag/blob/master/hoag/hoag.py
 
     Parameters