stock_k_means.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "     Stock  Stock.1  Stock.2  Stock.3  Stock.4  Stock.5  Stock.6  Stock.7  \\\n",
      "0    22650    81200   178500    32400    75500    41550    78500   146000   \n",
      "1    22550    82200   178500    32500    74500    40750    78800   147500   \n",
      "2    22600    82000   177500    32200    71800    40250    78900   145500   \n",
      "3    23900    86100   187500    33900    75000    42300    82300   152000   \n",
      "4    23200    86100   179500    34000    78100    42700    82700   151500   \n",
      "..     ...      ...      ...      ...      ...      ...      ...      ...   \n",
      "595  38600   148517   277000    38500    83100    31300    70800   130000   \n",
      "596  40800   152760   282000    40350    88600    31900    72700   134500   \n",
      "597  38500   153231   282000    41000    88400    31800    73900   137000   \n",
      "598  38850   153703   281500    41800    89600    31950    74100   137500   \n",
      "599  39150   154646   280500    40950    89900    31400    73200   136000   \n",
      "\n",
      "     Stock.8  Stock.9  ...  Stock.89  Stock.90  Stock.91  Stock.92  Stock.93  \\\n",
      "0      65500   155000  ...    104500    171000    126000    150000    233500   \n",
      "1      65900   146500  ...    101500    171500    122500    148000    235500   \n",
      "2      65900   145000  ...    102000    169000    123500    151000    236000   \n",
      "3      65700   154500  ...    104000    179500    126500    152500    250500   \n",
      "4      62200   154500  ...    100000    176000    129000    151000    249000   \n",
      "..       ...      ...  ...       ...       ...       ...       ...       ...   \n",
      "595    38383   120000  ...     95800    430000    161500    117500    417000   \n",
      "596    40032   126500  ...     99600    439500    165500    120000    431500   \n",
      "597    40673   131000  ...     97800    435500    163000    119000    416500   \n",
      "598    40581   128500  ...     91900    436000    162500    121500    414500   \n",
      "599    38932   128000  ...     92000    425500    164000    120500    420500   \n",
      "\n",
      "     Stock.94  Stock.95  Stock.96  Stock.97  Stock.98  \n",
      "0      152000     27050    820000     74700     28750  \n",
      "1      148000     27550    799000     75000     28650  \n",
      "2      148000     27950    794000     75400     28950  \n",
      "3      154000     27850    809000     80400     30600  \n",
      "4      148500     27000    802000     80400     33650  \n",
      "..        ...       ...       ...       ...       ...  \n",
      "595    143000     31600    481500     19450     52500  \n",
      "596    148000     32300    477500     20450     54000  \n",
      "597    149000     32150    460500     20350     54100  \n",
      "598    148000     32850    462000     20350     54700  \n",
      "599    146000     33100    450000     20350     55000  \n",
      "\n",
      "[600 rows x 99 columns] (600, 95)\n",
      "5 clusters for 40256 samples of data using 30 features\n",
      "Cluster 0 Summary: \n",
      "Cluster Size: 8747 Positive Growth: 50.771693 Average Growth: -0.014851\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Cluster 1 Summary: \n",
      "Cluster Size: 2405 Positive Growth: 45.031185 Average Growth: -2.008440\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXoAAAD8CAYAAAB5Pm/hAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjMuMywgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy/Il7ecAAAACXBIWXMAAAsTAAALEwEAmpwYAAAQUklEQVR4nO3df6zddX3H8edrrVSHEbRWp223dqHE1GEyuasu7oeRgcU56jKYRTO7jaSaSOZQo0UTwpjJZFvstowta8QF0aQQNuPNrOtQTJYYYb1FhRQE7/AHZTgLdDhmEIvv/XG+pMfjudxv6e29t5/7fCRNz/fz+Zx73uebk9f3cz7ne74nVYUkqV0/tdAFSJJOLINekhpn0EtS4wx6SWqcQS9JjTPoJalxvYI+yeYk9ySZTrJjTP+vJbk9yZEkF470bUvy9e7ftrkqXJLUT2Y7jz7JMuBe4FzgILAPuLiq7hoasw54HvBeYLKqburaXwBMARNAAfuBs6vq8Jw/E0nSWH1m9JuA6aq6r6qeAHYDW4YHVNU3q+oO4Ecj9309cHNVPdKF+83A5jmoW5LU0/IeY1YD9w9tHwRe1fPvj7vv6tFBSbYD2wFOPfXUs1/2spf1/POSJID9+/c/VFWrxvX1CfoTrqp2AbsAJiYmampqaoErkqSTS5JvzdTXZ+nmAWDt0Paarq2P47mvJGkO9An6fcCGJOuTnAJsBSZ7/v29wHlJnp/k+cB5XZskaZ7MGvRVdQS4lEFA3w3cWFUHklyV5AKAJL+U5CBwEfAPSQ50930E+FMGB4t9wFVdmyRpnsx6euV8c41eko5dkv1VNTGuz2/GSlLjDHpJapxBL0mNM+glqXEGvSQ1zqCXpMYZ9JLUOINekhpn0EtS4wx6SWqcQS9JjTPoJalxBr0kNc6gl6TGGfSS1DiDXpIaZ9BLUuMMeklqnEEvSY0z6CWpcQa9JDXOoJekxhn0ktQ4g16SGmfQS1LjDHpJapxBL0mNM+glqXEGvSQ1zqCXpMYZ9JLUOINekhpn0EtS4wx6SWqcQS9JjesV9Ek2J7knyXSSHWP6VyS5oeu/Lcm6rv1ZSa5LcmeSu5NcPsf1S5JmMWvQJ1kGXAOcD2wELk6ycWTYJcDhqjoD2Alc3bVfBKyoqrOAs4G3P3UQkCTNjz4z+k3AdFXdV1VPALuBLSNjtgDXdbdvAs5JEqCAU5MsB54DPAF8b04qlyT10ifoVwP3D20f7NrGjqmqI8CjwEoGof9/wIPAt4G/rKpHRh8gyfYkU0mmDh06dMxPQpI0sxP9Yewm4EngpcB64D1Jfn50UFXtqqqJqppYtWrVCS5JkpaWPkH/ALB2aHtN1zZ2TLdMcxrwMPAW4F+r6odV9V3gi8DE8RYtSeqvT9DvAzYkWZ/kFGArMDkyZhLY1t2+ELilqorBcs3rAJKcCrwa+NpcFC5J6mfWoO/W3C8F9gJ3AzdW1YEkVyW5oBt2LbAyyTTwbuCpUzCvAZ6b5ACDA8Y/VtUdc/0kJEkzy2DivXhMTEzU1NTUQpchSSeVJPurauzSuN+MlaTGGfSS1DiDXpIaZ9BLUuMMeklqnEEvSY0z6CWpccsXugDpZLHz5nvHtl927pnzXIl0bJzRS1LjDHpJapxBL0mNM+glqXEGvSQ1zqCXpMYZ9JLUOINekhpn0EtS4wx6SWqcQS9JjTPoJalxBr0kNc6gl6TGGfSS1DivRy8Nmema89LJzBm9JDXOoJekxhn0ktQ4g16SGmfQS1LjDHpJapynV2pJ8jRKLSXO6CWpcQa9JDXOoJekxvUK+iSbk9yTZDrJjjH9K5Lc0PXflmTdUN8rknwpyYEkdyZ59hzWL0maxaxBn2QZcA1wPrARuDjJxpFhlwCHq+oMYCdwdXff5cAngHdU1cuB1wI/nLPqJUmz6jOj3wRMV9V9VfUEsBvYMjJmC3Bdd/sm4JwkAc4D7qiqrwJU1cNV9eTclC5J6qNP0K8G7h/aPti1jR1TVUeAR4GVwJlAJdmb5PYk7xv3AEm2J5lKMnXo0KFjfQ6SpKdxoj+MXQ78CvDW7v/fTnLO6KCq2lVVE1U1sWrVqhNckiQtLX2C/gFg7dD2mq5t7JhuXf404GEGs/9/r6qHqur7wB7glcdbtCSpvz5Bvw/YkGR9klOArcDkyJhJYFt3+0LglqoqYC9wVpKf7g4Avw7cNTelS5L6mPUSCFV1JMmlDEJ7GfCxqjqQ5CpgqqomgWuB65NMA48wOBhQVYeTfITBwaKAPVX1mRP0XKQF8XSXU7js3DPnsRJpvF7XuqmqPQyWXYbbrhi6/Thw0Qz3/QSDUywlSQvAb8ZKUuMMeklqnEEvSY0z6CWpcQa9JDXOoJekxhn0ktQ4g16SGmfQS1LjDHpJapxBL0mNM+glqXEGvSQ1zqCXpMYZ9JLUOINekhpn0EtS4wx6SWqcQS9JjTPoJalxBr0kNc6gl6TGGfSS1DiDXpIaZ9BLUuMMeklqnEEvSY0z6CWpcQa9JDXOoJekxhn0ktQ4g16SGmfQS1LjDHpJapxBL0mN6xX0STYnuSfJdJIdY/pXJLmh678tybqR/p9N8liS985R3ZKknmYN+iTLgGuA84GNwMVJNo4MuwQ4XFVnADuBq0f6PwJ89vjLlSQdqz4z+k3AdFXdV1VPALuBLSNjtgDXdbdvAs5JEoAkbwK+ARyYk4olScekT9CvBu4f2j7YtY0dU1VHgEeBlUmeC7wf+JOne4Ak25NMJZk6dOhQ39olST2c6A9jrwR2VtVjTzeoqnZV1URVTaxateoElyRJS8vyHmMeANYOba/p2saNOZhkOXAa8DDwKuDCJH8OnA78KMnjVfW3x1u41MfOm+9d6BKkBdcn6PcBG5KsZxDoW4G3jIyZBLYBXwIuBG6pqgJ+9akBSa4EHjPktZTMdKC57Nwz57kSLWWzBn1VHUlyKbAXWAZ8rKoOJLkKmKqqSeBa4Pok08AjDA4GkqRFoM+MnqraA+wZabti6PbjwEWz/I0rn0F9kqTj5DdjJalxBr0kNc6gl6TGGfSS1DiDXpIaZ9BLUuMMeklqnEEvSY0z6CWpcQa9JDXOoJekxhn0ktQ4g16SGmfQS1LjDHpJapxBL0mNM+glqXEGvSQ1rtdPCUqaW/5ouOaTM3pJapxBL0mNM+glqXEGvSQ1zqCXpMYZ9JLUOINekhrnefQ66c10TrqkAWf0ktQ4g16SGmfQS1LjDHpJapxBL0mNM+glqXEGvSQ1rlfQJ9mc5J4k00l2jOlfkeSGrv+2JOu69nOT7E9yZ/f/6+a4fknSLGYN+iTLgGuA84GNwMVJNo4MuwQ4XFVnADuBq7v2h4DfqqqzgG3A9XNVuCSpnz4z+k3AdFXdV1VPALuBLSNjtgDXdbdvAs5Jkqr6clX9V9d+AHhOkhVzUbgkqZ8+Qb8auH9o+2DXNnZMVR0BHgVWjoz5HeD2qvrB6AMk2Z5kKsnUoUOH+tYuSephXj6MTfJyBss5bx/XX1W7qmqiqiZWrVo1HyVJ0pLRJ+gfANYOba/p2saOSbIcOA14uNteA3wKeFtV/efxFixJOjZ9gn4fsCHJ+iSnAFuByZExkww+bAW4ELilqirJ6cBngB1V9cU5qlmSdAxmDfpuzf1SYC9wN3BjVR1IclWSC7ph1wIrk0wD7waeOgXzUuAM4IokX+n+vWjOn4UkaUa9rkdfVXuAPSNtVwzdfhy4aMz9PgR86DhrlCQdB78ZK0mNM+glqXH+lKC0iDzdzyJedu6Z81iJWuKMXpIa54xeJw1/BFx6ZpzRS1LjnNFrUXHWLs09Z/SS1DiDXpIaZ9BLUuMMeklqnEEvSY0z6CWpcQa9JDXOoJekxhn0ktQ4vxmrE8YrMUqLgzN6SWqcM3rpJDHTOyTfHWk2zuglqXHO6HXcvOKktLg5o5ekxjmj14LwXYA0fwx69WY4Sycnl24kqXHO6KWTnF9M02wMeqlhnnsvcOlGkprnjF4/xg9cpfY4o5ekxhn0ktQ4l25OEnN9ZoVLNNLSYdA3wDMrdKw8JXNpMegb5qxdErhGL0nN6zWjT7IZ+GtgGfDRqvrwSP8K4OPA2cDDwJur6ptd3+XAJcCTwB9V1d45q34Wi/ntqcstWqx8bbZn1qBPsgy4BjgXOAjsSzJZVXcNDbsEOFxVZyTZClwNvDnJRmAr8HLgpcDnkpxZVU/O9RM50Z7JMogfkmqp86CxOKSqnn5A8svAlVX1+m77coCq+rOhMXu7MV9Kshz4DrAK2DE8dnjcTI83MTFRU1NTP9FuAEpLw0wHgWfyDn0pHWiS7K+qiXF9fZZuVgP3D20fBF4105iqOpLkUWBl137ryH1XjylwO7C923wsyT096jrRXgg8tNBFLBLui6PcF0edkH3x7nm4zzN5jFkshtfFz83UsSjOuqmqXcCuha5jWJKpmY6OS4374ij3xVHui6MW+77oc9bNA8Daoe01XdvYMd3SzWkMPpTtc19J0gnUJ+j3ARuSrE9yCoMPVydHxkwC27rbFwK31GDxfxLYmmRFkvXABuA/5qZ0SVIfsy7ddGvulwJ7GZxe+bGqOpDkKmCqqiaBa4Hrk0wDjzA4GNCNuxG4CzgCvPMkOuNmUS0lLTD3xVHui6PcF0ct6n0x61k3kqSTm9+MlaTGGfSS1DiDfowk70lSSV7YbSfJ3ySZTnJHklcudI0nWpK/SPK17vl+KsnpQ32Xd/viniSvX8Ay502Szd3znU6yY6HrmU9J1ib5QpK7khxI8q6u/QVJbk7y9e7/5y90rfMlybIkX07yL932+iS3da+PG7oTVxYNg35EkrXAecC3h5rPZ3DG0AYGX+z6+wUobb7dDPxCVb0CuBe4HGDkshabgb/rLpPRrKHLgJwPbAQu7vbDUnEEeE9VbQReDbyze/47gM9X1Qbg8932UvEu4O6h7auBnVV1BnCYwWVhFg2D/iftBN4HDH9KvQX4eA3cCpye5CULUt08qap/q6oj3eatDL4DAYN9sbuqflBV3wCmgU0LUeM82gRMV9V9VfUEsJvBflgSqurBqrq9u/2/DAJuNYN9cF037DrgTQtS4DxLsgb4TeCj3XaA1wE3dUMW3b4w6Ick2QI8UFVfHekadxmIn7iUQ8P+EPhsd3sp7oul+JzHSrIO+EXgNuDFVfVg1/Ud4MULVdc8+ysGk8Efddsrgf8ZmhgtutfHorgEwnxK8jngZ8Z0fRD4AINlmyXh6fZFVX26G/NBBm/dPzmftWnxSfJc4J+AP66q7w0msgNVVUmaP1c7yRuB71bV/iSvXeByeltyQV9VvzGuPclZwHrgq90LeA1we5JNNHoph5n2xVOS/D7wRuCcOvqFiyb3xSyW4nP+MUmexSDkP1lV/9w1/3eSl1TVg91S5ncXrsJ58xrggiRvAJ4NPI/Bb3WcnmR5N6tfdK8Pl246VXVnVb2oqtZV1ToGb79eWVXfYXAph7d1Z9+8Gnh06C1rk7ofm3kfcEFVfX+oayle1qLPZUCa1a1BXwvcXVUfGeoavvTJNuDT813bfKuqy6tqTZcRWxlc7uWtwBcYXP4FFuG+WHIz+mdoD/AGBh88fh/4g4UtZ178LbACuLl7h3NrVb3jJL+sxTMy02VAFris+fQa4PeAO5N8pWv7APBh4MYklwDfAn53YcpbFN4P7E7yIeDLDA6Mi4aXQJCkxrl0I0mNM+glqXEGvSQ1zqCXpMYZ9JLUOINekhpn0EtS4/4fQVKc5zHLNEoAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Cluster 2 Summary: \n",
      "Cluster Size: 2079 Positive Growth: 41.991342 Average Growth: -0.822450\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Cluster 3 Summary: \n",
      "Cluster Size: 18332 Positive Growth: 50.392756 Average Growth: -0.085527\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Cluster 4 Summary: \n",
      "Cluster Size: 8693 Positive Growth: 49.327045 Average Growth: -0.338963\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "5 clusters for 13419 samples of data using 30 features\n",
      "Fitting!\n",
      "Fitting!\n",
      "Fitting!\n",
      "Fitting!\n",
      "Fitting!\n",
      "No. Features: 30 No. Days: 5 No. Clusters: 5\n",
      "Cluster Number: 0 Cluster Size: 2869\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0       0.53      0.51      0.52      1392\n",
      "         1.0       0.55      0.57      0.56      1477\n",
      "\n",
      "    accuracy                           0.54      2869\n",
      "   macro avg       0.54      0.54      0.54      2869\n",
      "weighted avg       0.54      0.54      0.54      2869\n",
      "\n",
      "\n",
      "Cluster Number: 1 Cluster Size: 773\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0       0.60      0.85      0.70       432\n",
      "         1.0       0.59      0.26      0.36       341\n",
      "\n",
      "    accuracy                           0.59       773\n",
      "   macro avg       0.59      0.56      0.53       773\n",
      "weighted avg       0.59      0.59      0.55       773\n",
      "\n",
      "\n",
      "Cluster Number: 2 Cluster Size: 684\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0       0.63      0.75      0.69       383\n",
      "         1.0       0.58      0.44      0.50       301\n",
      "\n",
      "    accuracy                           0.61       684\n",
      "   macro avg       0.61      0.59      0.59       684\n",
      "weighted avg       0.61      0.61      0.60       684\n",
      "\n",
      "\n",
      "Cluster Number: 3 Cluster Size: 6139\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0       0.50      0.43      0.46      2974\n",
      "         1.0       0.53      0.61      0.56      3165\n",
      "\n",
      "    accuracy                           0.52      6139\n",
      "   macro avg       0.52      0.52      0.51      6139\n",
      "weighted avg       0.52      0.52      0.51      6139\n",
      "\n",
      "\n",
      "Cluster Number: 4 Cluster Size: 2954\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0       0.54      0.75      0.63      1540\n",
      "         1.0       0.53      0.30      0.39      1414\n",
      "\n",
      "    accuracy                           0.54      2954\n",
      "   macro avg       0.54      0.53      0.51      2954\n",
      "weighted avg       0.54      0.54      0.51      2954\n",
      "\n",
      "\n"
     ]
    }
   ],
   "source": [
    "import glob, os\n",
    "import pandas as pd\n",
    "import numpy as np\n",
    "import csv\n",
    "import matplotlib.pyplot as plt\n",
    "from sklearn.cluster import KMeans\n",
    "from sklearn import svm\n",
    "from sklearn.svm import SVC\n",
    "from sklearn import preprocessing\n",
    "from sklearn.model_selection import GridSearchCV\n",
    "from sklearn.metrics import classification_report\n",
    "from sklearn.model_selection import train_test_split\n",
    "\n",
    "'''\n",
    "Attempts to see if particular clutsering of price action results in a bias towards a particular \n",
    "price action. For example, certain candle formations are said to be associated with certain price \n",
    "action. eg. 'Descending/Ascending Triangle', 'wedge', trending up, trending down. etc.\n",
    "\n",
    "K-means clustering couldn't find significant bullish patterns, but Cluster 4 (downward trending)\n",
    "has an average growth rate of -1.9%\n",
    "\n",
    "Then, we use Support Vector Machine (SVM) to prediction price action movement. Best performance was \n",
    "Clutser 2 and 4, which had the following:\n",
    "\n",
    "Cluster Number: 2 Cluster Size: 674\n",
    "              precision    recall  f1-score   support\n",
    "\n",
    "         0.0       0.70      0.79      0.74       400\n",
    "         1.0       0.62      0.51      0.56       274\n",
    "\n",
    "    accuracy                           0.67       674\n",
    "   macro avg       0.66      0.65      0.65       674\n",
    "weighted avg       0.67      0.67      0.67       674\n",
    "\n",
    "Cluster Number: 4 Cluster Size: 809\n",
    "              precision    recall  f1-score   support\n",
    "\n",
    "         0.0       0.57      0.83      0.68       451\n",
    "         1.0       0.51      0.22      0.31       358\n",
    "\n",
    "    accuracy                           0.56       809\n",
    "   macro avg       0.54      0.53      0.49       809\n",
    "weighted avg       0.55      0.56      0.51       809\n",
    "\n",
    "\n",
    "\n",
    "'''\n",
    "\n",
    "\n",
    "def csv_extract(max, min, stock_num, filename = \"./KOSPI_DATA.csv\",\n",
    "                verbose = False, checknan = True):\n",
    "    '''\n",
    "    Input: Number of samples, Number of Stocks, filename\n",
    "    Output: numpy array of (number of samples, number of stocks)\n",
    "    '''\n",
    "    df = pd.read_csv(filename)\n",
    "    X = np.zeros((max-min, stock_num))\n",
    "    count = 0\n",
    "    for row in df:\n",
    "        if row[0] == 'S':\n",
    "            X[:, count] = df[row][min:max]\n",
    "            count = count + 1\n",
    "        if count == stock_num:\n",
    "            break\n",
    "    if checknan:\n",
    "        for i in range(X.shape[0]):\n",
    "            for j in range(X.shape[1]):\n",
    "                if np.isnan(X[i, j]):\n",
    "                    raise TypeError(\"Nan detected at indices \" + str((i, j)))\n",
    "    if verbose:\n",
    "        print(df, X.shape)\n",
    "    return X\n",
    "\n",
    "def data_extract(X_raw, train_max, feature_len, day_num):\n",
    "    X_train, y_train = np.empty((0, feature_len)), np.empty(0)\n",
    "    X_test, y_test = np.empty((0, feature_len)), np.empty(0)\n",
    "    for i in range(train_max):\n",
    "        for j in range(X_raw.shape[1]):\n",
    "            pct = 100*(X_raw[i + feature_len, j] - X_raw[i + feature_len + day_num - 1, j]\n",
    "                     ) / X_raw[i + feature_len, j]  \n",
    "            X_train = np.vstack((X_train, X_raw[i:i+feature_len, j]/np.max(X_raw[i:i+feature_len, j])))\n",
    "            y_train = np.append(y_train, pct)\n",
    "\n",
    "    for i in range(train_max, X_raw.shape[0] - day_num - feature_len):\n",
    "        for j in range(X_raw.shape[1]):\n",
    "            try:\n",
    "                pct = 100*(X_raw[i + feature_len, j] - X_raw[i + feature_len + day_num - 1, j]\n",
    "                     ) / X_raw[i + feature_len, j]\n",
    "            except Exception:\n",
    "                print(i, feature_len, day_num)\n",
    "            X_test = np.vstack((X_test, X_raw[i:i+feature_len, j]/np.max(X_raw[i:i+feature_len, j])))\n",
    "            y_test = np.append(y_test, pct)\n",
    "    return X_train, y_train, X_test, y_test\n",
    "\n",
    "class StockPredictor:\n",
    "    def __init__(self, X_train, X_test, y_train, y_test, cluster_num):\n",
    "        #X_train, X_test, y_train, y_test = train_test_split(X, y, random_state = np.random)\n",
    "        self.X_train, self.y_train = X_train, y_train\n",
    "        self.X_test, self.y_test = X_test, y_test\n",
    "        self.cluster_num = cluster_num\n",
    "  \n",
    "    def kmeans_train(self, verbose = True):\n",
    "        self.X_clust_train, self.y_clust_train = self.kmeans_fit(verbose = verbose,\n",
    "                                                             train_or_test = 'train')\n",
    "\n",
    "    def kmeans_test(self, verbose = True):\n",
    "        self.X_clust_test, self.y_clust_test = self.kmeans_fit(verbose = verbose, train_or_test = 'test')\n",
    "\n",
    "    def kmeans_fit(self, verbose = False, train_or_test = 'train'):\n",
    "        cluster_num = self.cluster_num\n",
    "        if train_or_test == 'train':\n",
    "            X, y = self.X_train, self.y_train\n",
    "            self.kmeans = KMeans(n_clusters = cluster_num, init = 'k-means++').fit(X)\n",
    "            labels = self.kmeans.labels_\n",
    "        elif train_or_test == 'test':\n",
    "            X, y = self.X_test, self.y_test\n",
    "            labels = self.kmeans.predict(X)\n",
    "        else:\n",
    "            print('Choose train or test. Returning None')\n",
    "            return None\n",
    "\n",
    "        X_clust, y_clust = {}, {}\n",
    "        for enum, label in enumerate(labels):\n",
    "            if label not in X_clust:\n",
    "                y_clust[label] = y[enum]\n",
    "                X_clust[label] = X[enum]\n",
    "                continue\n",
    "            y_clust[label] = np.append(y_clust[label], y[enum])\n",
    "            X_clust[label] = np.vstack((X_clust[label], X[enum]))    \n",
    "        if verbose:\n",
    "            print('%d clusters for %d samples of data using %d features'\n",
    "                %(cluster_num, enum+1, X.shape[1]))\n",
    "        return X_clust, y_clust\n",
    "\n",
    "    def svm_test(self, verbose = False):\n",
    "        self.y_clust_svm_test = self.get_svm_label(train_or_test = 'test')\n",
    "        self.X_clust_svm_test = self.get_svm_scaling(train_or_test = 'test')\n",
    "        for label in range(self.cluster_num):\n",
    "            clf = self.classifiers[label]\n",
    "            y_pred, y_true = clf.predict(self.X_clust_svm_test[label]), self.y_clust_svm_test[label]    \n",
    "            print('Cluster Number: %d Cluster Size: %d' %(label, y_pred.shape[0]))\n",
    "            print(classification_report(y_true, y_pred))\n",
    "            print()\n",
    "        return None\n",
    "        '''tot_right = 0; tot = 0\n",
    "        for label in range(self.cluster_num):\n",
    "          clf = self.classifiers[label]\n",
    "          y_predict = clf.predict(self.X_clust_svm_test[label])\n",
    "          if verbose:\n",
    "            print('CLUSTER %d:' %label)\n",
    "            print(np.sum(y_predict == self.y_clust_svm_test[label]), y_predict.shape[0], \n",
    "                100*np.sum(y_predict == self.y_clust_svm_test[label])/y_predict.shape[0])\n",
    "          tot_right = tot_right + np.sum(y_predict == self.y_clust_svm_test[label])\n",
    "          tot = tot + y_predict.shape[0]\n",
    "        if verbose:\n",
    "          print('Total Accuracy: %f' %(100*tot_right/tot))''\n",
    "        return tot_right/tot'''\n",
    "\n",
    "    def svm_fit(self):\n",
    "        self.y_clust_svm_train = self.get_svm_label(train_or_test = 'train')\n",
    "        self.X_clust_svm_train = self.get_svm_scaling(train_or_test = 'train')\n",
    "        self.classifiers = []\n",
    "        for key in range(self.cluster_num):\n",
    "            self.classifiers.append(self.get_svm_classifier(key))\n",
    "\n",
    "    def get_svm_classifier(self, key, verbose = False):\n",
    "        #clf = svm.SVC(kernel = 'rbf', gamma = 0.01, C = 100)\n",
    "        tuned_parameters = [{'kernel': ['rbf'], 'gamma': [1e-2, 1e-1, 1e+0, 1e+1],\n",
    "                         'C': [1, 10, 100, 1000]}]\n",
    "        tuned_parameters = [{'kernel': ['linear'], 'C': [1, 10, 100, 1000]}]\n",
    "        score = 'recall'\n",
    "        #clf = GridSearchCV(\n",
    "        #    SVC(), tuned_parameters, scoring='%s_macro' % score, verbose = 1\n",
    "        #)\n",
    "        #clf = SVC(kernel = 'poly', degree = 3, gamma = 1e-2, C = 100)\n",
    "        clf = SVC(kernel = 'linear', C = 10)\n",
    "        print('Fitting!')\n",
    "        clf.fit(self.X_clust_svm_train[key], self.y_clust_svm_train[key])\n",
    "        if verbose:\n",
    "            print(\"Best parameters set found on development set:\")\n",
    "            print(clf.best_params_)\n",
    "            print(\"Grid scores on development set:\")\n",
    "            means = clf.cv_results_['mean_test_score']\n",
    "            stds = clf.cv_results_['std_test_score']\n",
    "            for mean, std, params in zip(means, stds, clf.cv_results_['params']):\n",
    "                print(\"%0.3f (+/-%0.03f) for %r\"\n",
    "                      % (mean, std * 2, params))\n",
    "            print()\n",
    "        return clf\n",
    "\n",
    "    def get_svm_label(self, train_or_test = 'train'):\n",
    "        if train_or_test == 'train':\n",
    "            X_clust, y_clust = self.X_clust_train, self.y_clust_train\n",
    "        elif train_or_test == 'test':\n",
    "            X_clust, y_clust = self.X_clust_test, self.y_clust_test\n",
    "        else:\n",
    "            print('Choose train or test. Returning None')\n",
    "            return None\n",
    "\n",
    "        y_clust_svm = {}\n",
    "        for n in range(self.cluster_num):\n",
    "            y_clust_svm[n] = np.copy(y_clust[n])\n",
    "            for enum, i in enumerate(y_clust_svm[n]):\n",
    "                y_clust_svm[n][enum] = 0 if i < 0 else 1\n",
    "        return y_clust_svm\n",
    "\n",
    "    def get_svm_scaling(self, train_or_test = 'train'):\n",
    "        if train_or_test == 'train':\n",
    "            X_clust, keys = self.X_clust_train, self.X_clust_train.keys()\n",
    "        elif train_or_test == 'test':\n",
    "            X_clust, keys = self.X_clust_test, self.X_clust_test.keys()\n",
    "        else:\n",
    "            print('Choose train or test. Returning None')\n",
    "            return None\n",
    "\n",
    "        X_clust_svm = {}\n",
    "        for key in keys:\n",
    "            #X_clust_svm[key] = X_clust[key]\n",
    "            #continue\n",
    "            if train_or_test == 'train':\n",
    "                X_clust_svm[key] = preprocessing.scale(X_clust[key])\n",
    "            else:\n",
    "                shape = X_clust[key].shape\n",
    "                temp = np.vstack((X_clust[key], self.X_clust_train[key]))\n",
    "                X_clust_svm[key] = preprocessing.scale(temp)\n",
    "                X_clust_svm[key] = X_clust_svm[key][0:shape[0], 0:shape[1]]\n",
    "                #X_clust_svm[key] = X_clust[key]\n",
    "        return X_clust_svm\n",
    "\n",
    "    def evaluate_clusters(self, plotting = True):\n",
    "        X_clust = self.X_clust_train\n",
    "        y_clust = self.y_clust_train  \n",
    "        X_averages = []\n",
    "        for n in range(self.cluster_num):\n",
    "            pos = np.sum(y_clust[n] >= 0)\n",
    "            print('Cluster %d Summary: \\nCluster Size: %d Positive Growth: %f Average Growth: %f' \n",
    "            %(n, y_clust[n].shape[0], 100*pos/y_clust[n].shape[0], np.mean(y_clust[n])))\n",
    "            X_averages.append(np.mean(X_clust[n], axis=0))\n",
    "            if plotting:\n",
    "                plt.hist(y_clust[n], 50, density=True, alpha=0.5)\n",
    "                plt.xlim((-50, 50)); plt.ylim((0, 0.1))\n",
    "                plt.show()\n",
    "        if plotting:\n",
    "            for n in range(self.cluster_num):\n",
    "                plt.plot(X_averages[n])\n",
    "        plt.legend(labels = np.arange(0, self.cluster_num))        \n",
    "        plt.show()\n",
    "\n",
    "TRAIN_MAX = 300\n",
    "TEST_MAX = 600\n",
    "STOCK_NUM_TRAIN = 95\n",
    "STOCK_NUM_TEST = 10\n",
    "#(40, 7, 5)\n",
    "FEATURE_LEN = 30\n",
    "DAY_NUM = 5\n",
    "CLUSTER_NUM = 5\n",
    "\n",
    "SAMPLE_NUM_TEST = TEST_MAX-FEATURE_LEN-DAY_NUM\n",
    "SAMPLE_NUM_TRAIN = TRAIN_MAX-FEATURE_LEN-DAY_NUM\n",
    "\n",
    "if __name__ == \"__main__\":\n",
    "    filename = './DATA/KOSPI_DATA.csv'\n",
    "    X_raw = csv_extract(600, 0, STOCK_NUM_TRAIN, verbose = True, filename = filename)\n",
    "    X_train, y_train, X_test, y_test = data_extract(X_raw, TRAIN_MAX, FEATURE_LEN, DAY_NUM)\n",
    "\n",
    "    X_train, X_test, y_train, y_test = train_test_split(np.vstack((X_train, X_test)),\n",
    "                                                np.append(y_train, y_test), random_state = np.random)\n",
    "\n",
    "    model = StockPredictor(X_train, X_test, y_train, y_test, cluster_num = CLUSTER_NUM)\n",
    "    model.kmeans_train(verbose = True)\n",
    "    model.evaluate_clusters()\n",
    "    model.kmeans_test(verbose = True)\n",
    "    model.svm_fit()\n",
    "\n",
    "    print('No. Features: %d No. Days: %d No. Clusters: %d' %(FEATURE_LEN, DAY_NUM, CLUSTER_NUM))\n",
    "    model.svm_test(verbose = True)"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.9.0"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}