Merge pull request #6 from MetisMachine/th_hotfix

hotfix on data condition

Author: Tyler Hutcherson

bitcoinRNN.ipynb

@@ -3,7 +3,10 @@
   {
    "cell_type": "code",
    "execution_count": 1,
-   "metadata": {},
+   "metadata": {
+    "collapsed": false,
+    "scrolled": true
+   },
    "outputs": [
     {
      "name": "stdout",
@@ -35,7 +38,10 @@
   {
    "cell_type": "code",
    "execution_count": 2,
-   "metadata": {},
+   "metadata": {
+    "collapsed": false,
+    "scrolled": true
+   },
    "outputs": [
     {
      "name": "stdout",
@@ -77,7 +83,9 @@
   {
    "cell_type": "code",
    "execution_count": 18,
-   "metadata": {},
+   "metadata": {
+    "collapsed": true
+   },
    "outputs": [],
    "source": [
     "# # Bitcoin Price Prediction                                    \n",
@@ -118,7 +126,9 @@
   {
    "cell_type": "code",
    "execution_count": 163,
-   "metadata": {},
+   "metadata": {
+    "collapsed": false
+   },
    "outputs": [
     {
      "name": "stdout",
@@ -162,7 +172,9 @@
   {
    "cell_type": "code",
    "execution_count": 164,
-   "metadata": {},
+   "metadata": {
+    "collapsed": false
+   },
    "outputs": [
     {
      "name": "stdout",
@@ -184,7 +196,9 @@
   {
    "cell_type": "code",
    "execution_count": 165,
-   "metadata": {},
+   "metadata": {
+    "collapsed": false
+   },
    "outputs": [
     {
      "name": "stdout",
@@ -201,9 +215,11 @@
     "print(\"Validating a single record:\", flush=True)\n",
     "print(gtrends_json['data'][0], flush=True)\n",
     "\n",
-    "# Convert to pandas df\n",
+    "# Convert to pandas df and handle string NaN's coming through data engine\n",
     "gtrends = pd.DataFrame.from_records(gtrends_json['data'])\\\n",
-    "    .pivot(index='date', values='interest', columns='keyword')\n",
+    "    .pivot(index='date', values='interest', columns='keyword')\\\n",
+    "    .replace({'NaN': None})\\\n",
+    "    .fillna(method='pad')\n",
     "\n",
     "# Set proper date format\n",
     "gtrends.index = pd.to_datetime(gtrends.index)\n",
@@ -215,7 +231,9 @@
   {
    "cell_type": "code",
    "execution_count": 108,
-   "metadata": {},
+   "metadata": {
+    "collapsed": false
+   },
    "outputs": [
     {
      "name": "stdout",
@@ -245,7 +263,9 @@
   {
    "cell_type": "code",
    "execution_count": 109,
-   "metadata": {},
+   "metadata": {
+    "collapsed": true
+   },
    "outputs": [],
    "source": [
     "# Join google trends with quandl coin data\n",
@@ -292,7 +312,9 @@
   {
    "cell_type": "code",
    "execution_count": 113,
-   "metadata": {},
+   "metadata": {
+    "collapsed": true
+   },
    "outputs": [],
    "source": [
     "# ## Recurrent Neural Network Model\n",
@@ -317,7 +339,9 @@
   {
    "cell_type": "code",
    "execution_count": 140,
-   "metadata": {},
+   "metadata": {
+    "collapsed": false
+   },
    "outputs": [
     {
      "name": "stdout",
@@ -444,7 +468,9 @@
   {
    "cell_type": "code",
    "execution_count": 141,
-   "metadata": {},
+   "metadata": {
+    "collapsed": true
+   },
    "outputs": [],
    "source": [
     "# Predict over the holdout test set and retain the hidden state\n",
@@ -462,7 +488,9 @@
   {
    "cell_type": "code",
    "execution_count": 142,
-   "metadata": {},
+   "metadata": {
+    "collapsed": true
+   },
    "outputs": [],
    "source": [
     "# Get the prediction and date value\n",
@@ -473,7 +501,9 @@
   {
    "cell_type": "code",
    "execution_count": 143,
-   "metadata": {},
+   "metadata": {
+    "collapsed": false
+   },
    "outputs": [
     {
      "name": "stdout",
@@ -492,7 +522,9 @@
   {
    "cell_type": "code",
    "execution_count": 160,
-   "metadata": {},
+   "metadata": {
+    "collapsed": true
+   },
    "outputs": [],
    "source": [
     "data_out = [{'price_prediction': predicted_price,\n",
@@ -504,7 +536,9 @@
   {
    "cell_type": "code",
    "execution_count": 161,
-   "metadata": {},
+   "metadata": {
+    "collapsed": true
+   },
    "outputs": [],
    "source": [
     "# # Persist Predictions\n",
@@ -528,7 +562,9 @@
   {
    "cell_type": "code",
    "execution_count": 162,
-   "metadata": {},
+   "metadata": {
+    "collapsed": false
+   },
    "outputs": [
     {
      "name": "stdout",
@@ -550,10 +586,11 @@
   }
  ],
  "metadata": {
+  "anaconda-cloud": {},
   "kernelspec": {
-   "display_name": "Python 3",
+   "display_name": "Python [Root]",
    "language": "python",
-   "name": "python3"
+   "name": "Python [Root]"
   },
   "language_info": {
    "codemirror_mode": {
@@ -565,7 +602,7 @@
    "name": "python",
    "nbconvert_exporter": "python",
    "pygments_lexer": "ipython3",
-   "version": "3.6.2"
+   "version": "3.5.4"
   }
  },
  "nbformat": 4,

main.py

@@ -48,8 +48,11 @@
 print("Validating a single record:", flush=True)
 print(gtrends_json['data'][0], flush=True)
 
-# Convert to pandas df
-gtrends = pd.DataFrame.from_records(gtrends_json['data'])    .pivot(index='date', values='interest', columns='keyword')
+# Convert to pandas df and handles string NaN's coming through data engine
+gtrends = pd.DataFrame.from_records(gtrends_json['data'])\
+    .pivot(index='date', values='interest', columns='keyword')\
+    .replace({'NaN': None})\
+    .fillna(method='pad')
 
 # Set proper date format
 gtrends.index = pd.to_datetime(gtrends.index)