Merge pull request #126 from DoubleML/s-guide-rework

Rework the Guide and add Release Notes for 0.6.0

Author: SvenKlaassen

doc/api/api.rst

@@ -30,6 +30,7 @@ Double machine learning models
     DoubleMLLPQ
     DoubleMLCVAR
     DoubleMLQTE
+    DoubleMLBLP
 
 Datasets module
 ---------------

doc/examples/index.rst

@@ -1,6 +1,8 @@
 
 :parenttoc: True
 
+.. _examplegallery:
+
 Examples
 ==========
 
@@ -30,6 +32,7 @@ These are case studies with the Python package :ref:`DoubleML <doubleml_package>
     py_double_ml_pension_qte.ipynb
     py_double_ml_pq.ipynb
     py_double_ml_cvar.ipynb
+    py_double_ml_learner.ipynb
 
 |start-h3| Sandbox |end-h3|
 

doc/examples/py_double_ml_cate.ipynb

@@ -12,14 +12,15 @@
    ]
   },
   {
+   "attachments": {},
    "cell_type": "markdown",
    "metadata": {
     "collapsed": false
    },
    "source": [
     "## Data\n",
     "\n",
-    "We define a data generating process to create synthetic data to compare the estimates to the true effect. The data generating process is based on the Monte Carlo simulation from this [paper](https://arxiv.org/abs/1806.03467) and this implementation from [EconML](https://github.com/microsoft/EconML)."
+    "We define a data generating process to create synthetic data to compare the estimates to the true effect. The data generating process is based on the Monte Carlo simulation from [Oprescu et al. (2019)](http://proceedings.mlr.press/v97/oprescu19a.html) and this [notebook](https://github.com/py-why/EconML/blob/main/notebooks/Causal%20Forest%20and%20Orthogonal%20Random%20Forest%20Examples.ipynb) from [EconML](https://github.com/py-why/EconML)."
    ]
   },
   {
@@ -80,33 +81,6 @@
     "    return te\n",
     "\n",
     "def create_synthetic_data(n_samples=200, n_w=30, support_size=5, n_x=1):\n",
-    "    \"\"\"\n",
-    "    Creates a simple synthetic example for conditional treatment effects.\n",
-    "\n",
-    "    Parameters\n",
-    "    ----------\n",
-    "    n_samples : int\n",
-    "        Number of samples.\n",
-    "        Default is ``200``.\n",
-    "\n",
-    "    n_w : int\n",
-    "        Dimension of covariates.\n",
-    "        Default is ``30``.\n",
-    "\n",
-    "    support_size : int\n",
-    "        Number of relevant covariates.\n",
-    "        Default is ``5``.\n",
-    "\n",
-    "    n_x : int\n",
-    "        Dimension of treatment variable.\n",
-    "        Default is ``1``.\n",
-    "\n",
-    "    Returns\n",
-    "    -------\n",
-    "     data : pd.DataFrame\n",
-    "            A data frame.\n",
-    "\n",
-    "    \"\"\"\n",
     "    # Outcome support\n",
     "    # With the next two lines we are effectively choosing the matrix gamma in the example\n",
     "    support_y = np.random.choice(np.arange(n_w), size=support_size, replace=False)\n",
@@ -219,12 +193,13 @@
    ]
   },
   {
+   "attachments": {},
    "cell_type": "markdown",
    "metadata": {
     "collapsed": false
    },
    "source": [
-    "To estimate the CATE, we rely on the best-linear-predictor of the linear score as in [Semenova et al.](https://doi.org/10.1093/ectj/utaa027) To approximate the target function $g(x)$ with a linear form, we have to define a data frame of basis functions. Here, we rely on [patsy](https://patsy.readthedocs.io/en/latest/) to construct a suitable basis of [B-splines](https://en.wikipedia.org/wiki/B-spline)."
+    "To estimate the CATE, we rely on the best-linear-predictor of the linear score as in [Semenova et al. (2021)](https://doi.org/10.1093/ectj/utaa027) To approximate the target function $g(x)$ with a linear form, we have to define a data frame of basis functions. Here, we rely on [patsy](https://patsy.readthedocs.io/en/latest/) to construct a suitable basis of [B-splines](https://en.wikipedia.org/wiki/B-spline)."
    ]
   },
   {
@@ -241,6 +216,7 @@
    ]
   },
   {
+   "attachments": {},
    "cell_type": "markdown",
    "metadata": {
     "collapsed": false
@@ -262,6 +238,7 @@
    ]
   },
   {
+   "attachments": {},
    "cell_type": "markdown",
    "metadata": {
     "collapsed": false