patch for object dtype in MonteCarloShapleyBatch

Author: simon.grah

nbs/monte_carlo_shapley.ipynb

@@ -302,8 +302,9 @@
     "\n",
     "    # Compute the matrix X of hybrid individuals between x and ref\n",
     "    # and keep trace of permutation orders\n",
-    "    array_of_hybrid_individuals = np.zeros(shape=(n_iter * (dimension+1), dimension))\n",
-    "    orders = np.zeros(shape=(n_iter * (dimension+1),), dtype=int)\n",
+    "    dtype_features = np.object if x.dtype == np.object else np.float\n",
+    "    array_of_hybrid_individuals = np.empty(shape=(n_iter * (dimension+1), dimension), dtype=dtype_features)\n",
+    "    orders = np.zeros(shape=(n_iter * (dimension+1),), dtype=np.int32)\n",
     "\n",
     "    for iter_monte_carlo in range(n_iter):\n",
     "        order = np.random.permutation(dimension)\n",
@@ -462,7 +463,7 @@
      "name": "stderr",
      "output_type": "stream",
      "text": [
-      "100%|██████████| 10000/10000 [00:00<00:00, 12426.16it/s]\n"
+      "100%|██████████| 10000/10000 [00:00<00:00, 12464.78it/s]\n"
      ]
     }
    ],
@@ -687,7 +688,7 @@
      "name": "stderr",
      "output_type": "stream",
      "text": [
-      "100%|██████████| 10000/10000 [00:01<00:00, 7932.50it/s]\n"
+      "100%|██████████| 10000/10000 [00:01<00:00, 6035.08it/s]\n"
      ]
     }
    ],
@@ -830,7 +831,7 @@
      "name": "stderr",
      "output_type": "stream",
      "text": [
-      "100%|██████████| 5/5 [00:00<00:00, 675.48it/s]\n"
+      "100%|██████████| 5/5 [00:00<00:00, 648.77it/s]\n"
      ]
     }
    ],
@@ -857,7 +858,7 @@
      "name": "stderr",
      "output_type": "stream",
      "text": [
-      "100%|██████████| 5/5 [00:00<00:00, 54.69it/s]\n"
+      "100%|██████████| 5/5 [00:00<00:00, 55.76it/s]\n"
      ]
     }
    ],

nbs/plots.ipynb

@@ -8,7 +8,7 @@ author = Simon Grah
 author_email = simon.grah.pro@gmail.com
 copyright = THALES SIX GTS France
 branch = master
-version = 0.0.2
+version = 0.0.3
 min_python = 3.6
 audience = Developers
 language = English

settings.ini

@@ -1 +1 @@
-__version__ = "0.0.2"
+__version__ = "0.0.3"

shapkit/__init__.py

@@ -138,8 +138,9 @@ def MonteCarloShapleyBatch(x, fc, ref, n_iter):
 
     # Compute the matrix X of hybrid individuals between x and ref
     # and keep trace of permutation orders
-    array_of_hybrid_individuals = np.zeros(shape=(n_iter * (dimension+1), dimension))
-    orders = np.zeros(shape=(n_iter * (dimension+1),), dtype=int)
+    dtype_features = np.object if x.dtype == np.object else np.float
+    array_of_hybrid_individuals = np.empty(shape=(n_iter * (dimension+1), dimension), dtype=dtype_features)
+    orders = np.zeros(shape=(n_iter * (dimension+1),), dtype=np.int32)
 
     for iter_monte_carlo in range(n_iter):
         order = np.random.permutation(dimension)

shapkit/monte_carlo_shapley.py

@@ -37,7 +37,7 @@
 sns.set_style("whitegrid")
 
 # Cell
-def plot_shapley(x, fc, ref, shapley_values, n_attributes, savefig=False, filename="fig.png", fig_format="png"):
+def plot_shapley(x, fc, ref, shapley_values, n_attributes, savefig=False, filename="fig.png", fig_format="png", no_show=False):
 
     # Get output for x and ref
     pred_x = fc(x.values)
@@ -109,4 +109,7 @@ def plot_shapley(x, fc, ref, shapley_values, n_attributes, savefig=False, filena
     if savefig == True:
         plt.savefig(filename, format=fig_format, bbox_inches='tight')
         print("plot saved with path {0} and {1} format".format(filename, fig_format))
-    plt.show()
+    if no_show == True:
+        return fig
+    else:
+        plt.show()