Update .ipynb and .py for DL - plot classification error, not MSE

arnocandel · arnocandel · commit 2eb7e8e3e4ed · 2015-11-07T13:22:48.000-08:00
diff --git a/tutorials/deeplearning/deeplearning.ipynb b/tutorials/deeplearning/deeplearning.ipynb
@@ -489,7 +489,7 @@
    "cell_type": "code",
    "execution_count": null,
    "metadata": {
-    "collapsed": true
+    "collapsed": false
    },
    "outputs": [],
    "source": [
@@ -526,7 +526,7 @@
    },
    "outputs": [],
    "source": [
-    "var_df = pd.DataFrame(covtype_model_v1.varimp(return_list=True), \n",
+    "var_df = pd.DataFrame(covtype_model_v1.varimp(), \n",
     "             columns=[\"Variable\", \"Relative Importance\", \"Scaled Importance\", \"Percentage\"])\n",
     "print var_df.shape\n",
     "var_df.head(10)"
@@ -574,7 +574,7 @@
    "cell_type": "code",
    "execution_count": null,
    "metadata": {
-    "collapsed": true
+    "collapsed": false
    },
    "outputs": [],
    "source": [
@@ -590,8 +590,8 @@
    },
    "outputs": [],
    "source": [
-    "plt.plot(cov_v2_df['training_MSE'], label=\"training_mse\")\n",
-    "plt.plot(cov_v2_df['validation_MSE'], label=\"validation_mse\")\n",
+    "plt.plot(cov_v2_df['training_classification_error'], label=\"training_classification_error\")\n",
+    "plt.plot(cov_v2_df['validation_classification_error'], label=\"validation_classification_error\")\n",
     "plt.title(\"Covertype Deep Learner (Early Stop)\")\n",
     "plt.legend();"
    ]
@@ -627,17 +627,17 @@
     "covtype_model_tuned = H2ODeepLearningEstimator(\n",
     "  model_id=\"covtype_tuned\",\n",
     "  overwrite_with_best_model=False,\n",
-    "  hidden=[128,128,128],           # more hidden layers -> more complex interactions\n",
+    "  hidden=[128,128,128],            # more hidden layers -> more complex interactions\n",
     "  epochs=10,                       # to keep it short enough\n",
     "  score_validation_samples=10000,  # downsample validation set for faster scoring\n",
     "  score_duty_cycle=0.025,          # don't score more than 2.5% of the wall time\n",
-    "  adaptive_rate=False,                 # manually tuned learning rate\n",
+    "  adaptive_rate=False,             # manually tuned learning rate\n",
     "  rate=0.01, \n",
     "  rate_annealing=0.000002,            \n",
     "  momentum_start=0.2,              # manually tuned momentum\n",
     "  momentum_stable=0.4, \n",
     "  momentum_ramp=10000000, \n",
-    "  l1=0.00001,                         # add some L1/L2 regularization\n",
+    "  l1=0.00001,                      # add some L1/L2 regularization\n",
     "  l2=0.00001,\n",
     "  max_w2=10                        # helps stability for Rectifier\n",
     ")\n",
@@ -655,7 +655,7 @@
    "cell_type": "code",
    "execution_count": null,
    "metadata": {
-    "collapsed": true
+    "collapsed": false
    },
    "outputs": [],
    "source": [
@@ -672,8 +672,8 @@
    },
    "outputs": [],
    "source": [
-    "plt.plot(cov_tuned_df['training_MSE'], label=\"training_mse\")\n",
-    "plt.plot(cov_tuned_df['validation_MSE'], label=\"validation_mse\")\n",
+    "plt.plot(cov_tuned_df['training_classification_error'], label=\"training_classification_error\")\n",
+    "plt.plot(cov_tuned_df['validation_classification_error'], label=\"validation_classification_error\")\n",
     "plt.title(\"Covertype Deep Learner (Tuned)\")\n",
     "plt.legend();"
    ]
diff --git a/tutorials/deeplearning/deeplearning.py b/tutorials/deeplearning/deeplearning.py
@@ -296,7 +296,7 @@ def plot_spirals(models, model_names):
 
 # In[ ]:
 
-var_df = pd.DataFrame(covtype_model_v1.varimp(return_list=True), 
+var_df = pd.DataFrame(covtype_model_v1.varimp(), 
              columns=["Variable", "Relative Importance", "Scaled Importance", "Percentage"])
 print var_df.shape
 var_df.head(10)
@@ -332,8 +332,8 @@ def plot_spirals(models, model_names):
 
 # In[ ]:
 
-plt.plot(cov_v2_df['training_MSE'], label="training_mse")
-plt.plot(cov_v2_df['validation_MSE'], label="validation_mse")
+plt.plot(cov_v2_df['training_classification_error'], label="training_classification_error")
+plt.plot(cov_v2_df['validation_classification_error'], label="validation_classification_error")
 plt.title("Covertype Deep Learner (Early Stop)")
 plt.legend();
 
@@ -353,17 +353,17 @@ def plot_spirals(models, model_names):
 covtype_model_tuned = H2ODeepLearningEstimator(
   model_id="covtype_tuned",
   overwrite_with_best_model=False,
-  hidden=[128,128,128],           # more hidden layers -> more complex interactions
+  hidden=[128,128,128],            # more hidden layers -> more complex interactions
   epochs=10,                       # to keep it short enough
   score_validation_samples=10000,  # downsample validation set for faster scoring
   score_duty_cycle=0.025,          # don't score more than 2.5% of the wall time
-  adaptive_rate=False,                 # manually tuned learning rate
+  adaptive_rate=False,             # manually tuned learning rate
   rate=0.01, 
   rate_annealing=0.000002,            
   momentum_start=0.2,              # manually tuned momentum
   momentum_stable=0.4, 
   momentum_ramp=10000000, 
-  l1=0.00001,                         # add some L1/L2 regularization
+  l1=0.00001,                      # add some L1/L2 regularization
   l2=0.00001,
   max_w2=10                        # helps stability for Rectifier
 )
@@ -380,8 +380,8 @@ def plot_spirals(models, model_names):
 
 # In[ ]:
 
-plt.plot(cov_tuned_df['training_MSE'], label="training_mse")
-plt.plot(cov_tuned_df['validation_MSE'], label="validation_mse")
+plt.plot(cov_tuned_df['training_classification_error'], label="training_classification_error")
+plt.plot(cov_tuned_df['validation_classification_error'], label="validation_classification_error")
 plt.title("Covertype Deep Learner (Tuned)")
 plt.legend();
 
