AUDJPY_gbc.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "import pandas as pd\n",
    "\n",
    "import matplotlib.pyplot as plt\n",
    "from sklearn import ensemble"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "head = ['HK Time', 'Open', 'High', 'Low', 'Close', 'Volume']\n",
    "data_2017 = pd.read_csv(\"F:\\personal project\\iq option trade\\OUT_DAT_ASCII_AUDJPY_M1_2017.csv\", delimiter=\";\", \n",
    "                        names=head)\n",
    "data_2018 = pd.read_csv(\"F:\\personal project\\iq option trade\\OUT_DAT_ASCII_AUDJPY_M1_2018.csv\", delimiter=\";\",\n",
    "                        names=head)\n",
    "data_2019 = pd.read_csv(\"F:\\personal project\\iq option trade\\OUT_DAT_ASCII_AUDJPY_M1_201901.csv\", delimiter=\";\",\n",
    "                        names=head)\n",
    "train_data = data_2017.append(data_2018, ignore_index=True)\n",
    "train_data = train_data.drop('Volume', 1)\n",
    "data_2019 = data_2019.drop('Volume', 1)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Create return (open-close) normalized\n",
    "train_data['Close'].plot()\n",
    "train_data['return_next'] = train_data['Open'].shift(-1) - train_data['Close'].shift(-1)\n",
    "return_range = train_data['return_next'].max() - train_data['return_next'].min()\n",
    "train_data['return_next'] = train_data['return_next'] / return_range\n",
    "\n",
    "train_data['return'] = train_data['Open'] - train_data['Close']\n",
    "return_range = train_data['return'].max() - train_data['return'].min()\n",
    "train_data['return'] = train_data['return'] / return_range\n",
    "\n",
    "# Make label, 1 as rising price, 0 as falling price - prediction for the following timestamp\n",
    "train_data['label'] = train_data['return_next'].apply(lambda x: 1 if x > 0.0 else 0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Create return (open-close) normalized\n",
    "data_2019['Close'].plot()\n",
    "data_2019['return_next'] = data_2019['Open'].shift(-1) - data_2019['Close'].shift(-1)\n",
    "return_range = data_2019['return_next'].max() - data_2019['return_next'].min()\n",
    "data_2019['return_next'] = data_2019['return_next'] / return_range\n",
    "\n",
    "data_2019['return'] = data_2019['Open'] - data_2019['Close']\n",
    "return_range = data_2019['return'].max() - data_2019['return'].min()\n",
    "data_2019['return'] = data_2019['return'] / return_range\n",
    "\n",
    "# If return > 0, green; otherwise, red\n",
    "data_2019['return'].plot()\n",
    "# Make label, 1 as rising price, 0 as falling price - prediction for the following timestamp\n",
    "data_2019['label'] = data_2019['return_next'].apply(lambda x: 1 if x > 0.0 else 0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(25900, 8)\n(25899, 14) Index(['HK Time', 'Open', 'High', 'Low', 'Close', 'return_next', 'return',\n       'label', 'datetime', 'year', 'month', 'day', 'hour', 'minute'],\n      dtype='object')\n"
     ]
    }
   ],
   "source": [
    "print(data_2019.shape)\n",
    "data_2019.dropna(inplace=True)\n",
    "data_2019['datetime'] = pd.to_datetime(data_2019['HK Time'])\n",
    "data_2019 = data_2019.set_index(data_2019['datetime'])\n",
    "data_2019.index.names = [None]\n",
    "data_2019['year'] = data_2019.index.year\n",
    "data_2019['month'] = data_2019.index.month\n",
    "data_2019['day'] = data_2019.index.day\n",
    "data_2019['hour'] = data_2019.index.hour\n",
    "data_2019['minute'] = data_2019.index.minute\n",
    "print(data_2019.shape, data_2019.columns)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(745007, 8)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(745006, 14) Index(['HK Time', 'Open', 'High', 'Low', 'Close', 'return_next', 'return',\n       'label', 'datetime', 'year', 'month', 'day', 'hour', 'minute'],\n      dtype='object')\n"
     ]
    }
   ],
   "source": [
    "print(train_data.shape)\n",
    "train_data.dropna(inplace=True)\n",
    "train_data['datetime'] = pd.to_datetime(train_data['HK Time'])\n",
    "train_data = train_data.set_index(train_data['datetime'])\n",
    "train_data.index.names = [None]\n",
    "train_data['year'] = train_data.index.year\n",
    "train_data['month'] = train_data.index.month\n",
    "train_data['day'] = train_data.index.day\n",
    "train_data['hour'] = train_data.index.hour\n",
    "train_data['minute'] = train_data.index.minute\n",
    "print(train_data.shape, train_data.columns)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "df_test = data_2019\n",
    "df_train = train_data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Separate features and output\n",
    "\n",
    "y_train = df_train['label']\n",
    "x_train = df_train.loc[:, ['Open', 'High', 'Low', 'Close', 'return', 'year', 'month', 'day', 'hour', 'minute']]\n",
    "\n",
    "y_test = df_test['label']\n",
    "x_test = df_test.loc[:, ['Open', 'High', 'Low', 'Close', 'return', 'year', 'month', 'day', 'hour', 'minute']]\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(25899, 10)\n(25899,)\n% of Class0 : 0.521371 y_test\n% of Class1 : 0.478629 y_test\n"
     ]
    }
   ],
   "source": [
    "print(x_test.shape)\n",
    "print(y_test.shape)\n",
    "\n",
    "print('%% of Class0 : %f y_test' % (np.count_nonzero(y_test == 0)/float(len(y_test))))\n",
    "print('%% of Class1 : %f y_test' % (np.count_nonzero(y_test == 1)/float(len(y_test))))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(745006, 10)\n(745006,)\n% of Class0 : 0.525080 y_train\n% of Class1 : 0.474920 y_train\n"
     ]
    }
   ],
   "source": [
    "print(x_train.shape)\n",
    "print(y_train.shape)\n",
    "\n",
    "print('%% of Class0 : %f y_train' % (np.count_nonzero(y_train == 0)/float(len(y_train))))\n",
    "print('%% of Class1 : %f y_train' % (np.count_nonzero(y_train == 1)/float(len(y_train))))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train Accuracy: 0.530441\n"
     ]
    }
   ],
   "source": [
    "# Define GB Classifier Model and fit\n",
    "\n",
    "gbc = ensemble.GradientBoostingClassifier()\n",
    "gbc.fit(x_train, y_train)\n",
    "accuracy = gbc.score(x_train, y_train)\n",
    "print('Train Accuracy: %f' % accuracy)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Put in test data\n",
    "df_trade = x_test.copy()\n",
    "df_trade['label'] = y_test"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.5194022935248466\n"
     ]
    }
   ],
   "source": [
    "# Predict GBC with test data\n",
    "pred_gbc = gbc.predict(x_test)\n",
    "df_trade['pred'] = pred_gbc\n",
    "df_trade['won'] = df_trade['label'] == df_trade['pred']\n",
    "df_trade.head(10)\n",
    "print(df_trade[df_trade['won'] == True]['won'].count()/df_trade['won'].count())\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 2",
   "language": "python",
   "name": "python2"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 2
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython2",
   "version": "2.7.6"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 0
}