revise source code

Author: Cheng-Lin-Li

Competition/McKinseyAnalyticsPrediction/NumberOfVehiclesPrediction.ipynb

@@ -759,7 +759,8 @@
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "## Build model: Try NN, LSTM, GRU for this prediction"
+    "## Build model: Try NN, LSTM, GRU for this prediction\n",
+    "### Construct a 4 layers full connected deep learning neural network to perform prediction by early stop."
    ]
   },
   {
@@ -972,7 +973,9 @@
     "for i in range(100):\n",
     "\tmodel.fit(trainX, trainY, epochs=1, batch_size=batch_size, verbose=2, shuffle=False)\n",
     "\tmodel.reset_states()\n",
-    "```"
+    "```\n",
+    "\n",
+    "### Construct a 4 layers deep learning LSTM network to perform prediction."
    ]
   },
   {
@@ -1180,7 +1183,9 @@
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "## GRU model"
+    "## Gated Recurrent Unit (GRU) model\n",
+    "\n",
+    "### Construct a 4 layers deep learning GRU network to perform prediction."
    ]
   },
   {
@@ -1356,11 +1361,18 @@
    ]
   },
   {
-   "cell_type": "code",
-   "execution_count": null,
+   "cell_type": "markdown",
    "metadata": {},
-   "outputs": [],
-   "source": []
+   "source": [
+    "## GRU get better result in similar parameter settings."
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "## All those parameters still have a lot of space to fine-tune."
+   ]
   },
   {
    "cell_type": "code",