GaussianProcessRegression_basic.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1114,
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "import numpy as np\n",
    "from matplotlib import pyplot as plt\n",
    "from sklearn.gaussian_process import GaussianProcessRegressor\n",
    "from sklearn.gaussian_process.kernels import (RBF, Matern, RationalQuadratic,\n",
    "                                              ExpSineSquared, DotProduct,\n",
    "                                              ConstantKernel as C)\n",
    "import statsmodels\n",
    "from statsmodels import \n",
    "from sklearn.metrics import mean_absolute_error"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1116,
   "metadata": {},
   "outputs": [
    {
     "ename": "ImportError",
     "evalue": "cannot import name 'SimpleExpSmoothing'",
     "output_type": "error",
     "traceback": [
      "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[0;31mImportError\u001b[0m                               Traceback (most recent call last)",
      "\u001b[0;32m<ipython-input-1116-886bd1cc9a16>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m()\u001b[0m\n\u001b[1;32m      1\u001b[0m \u001b[0;32mimport\u001b[0m \u001b[0mstatsmodels\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mapi\u001b[0m \u001b[0;32mas\u001b[0m \u001b[0msm\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      2\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m----> 3\u001b[0;31m \u001b[0;32mfrom\u001b[0m \u001b[0mstatsmodels\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mtsa\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mapi\u001b[0m \u001b[0;32mimport\u001b[0m \u001b[0mSimpleExpSmoothing\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m",
      "\u001b[0;31mImportError\u001b[0m: cannot import name 'SimpleExpSmoothing'"
     ]
    }
   ],
   "source": [
    "import statsmodels.api as sm\n",
    "\n",
    "from statsmodels.tsa.api import SimpleExpSmoothing "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "statsmodels."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1096,
   "metadata": {},
   "outputs": [],
   "source": [
    "df=pd.read_csv('~/Desktop/ML/periodic.csv')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1097,
   "metadata": {},
   "outputs": [],
   "source": [
    "X_df=df[\"t\"] \n",
    "y_df=df[\"Sum\"] "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1098,
   "metadata": {},
   "outputs": [],
   "source": [
    "X_train=np.array(  [X_df.iloc[:28]]  ).T\n",
    "y_train=np.array([y_df.iloc[:28] ]).T\n",
    "\n",
    "X_test=np.array([X_df.iloc[28:]]).T\n",
    "y_test=np.array([y_df.iloc[28:]]).T\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1099,
   "metadata": {},
   "outputs": [],
   "source": [
    "#ExpSine for periodic\n",
    "##Smooth rising trend is to be explained by an RBF kernel\n",
    "#The RBF kernel with a large length-scale enforces this component to be smooth; \n",
    "#it is not enforced that the trend is rising which leaves this choice to the GP\n",
    "\n",
    "kernels = [   \n",
    "    1.0* RBF(length_scale=500) + \n",
    "    1.0* RBF(length_scale=50) *\n",
    "    1.0* ExpSineSquared(length_scale=1.0, periodicity=7.0,length_scale_bounds=(10, 100),periodicity_bounds=(7, 7) ) \n",
    "      +    \n",
    "    1.0* RBF(length_scale=50) *\n",
    "    1.0*ExpSineSquared(length_scale=1.0, periodicity=7.0,length_scale_bounds=(0.1, 10.0),periodicity_bounds=(7, 7) )\n",
    "    ,\n",
    "    \n",
    "    1.0*Matern(length_scale=1.0, length_scale_bounds=(1, 1000), nu=5)+\n",
    "    1.0*Matern(length_scale=1.0, length_scale_bounds=(1, 1000), nu=5)*   \n",
    "      1.0* ExpSineSquared(length_scale=1.0, periodicity=7.0,length_scale_bounds=(1, 100),periodicity_bounds=(7, 7))\n",
    "    +\n",
    "   \n",
    "      1.0* ExpSineSquared(length_scale=1.0, periodicity=7.0,length_scale_bounds=(1, 100),periodicity_bounds=(7, 7))\n",
    "]\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1100,
   "metadata": {},
   "outputs": [],
   "source": [
    "n_restarts_optimizer=10\n",
    "gp1 = GaussianProcessRegressor(kernel=kernels[0], n_restarts_optimizer=n_restarts_optimizer)\n",
    "gp2 = GaussianProcessRegressor(kernel=kernels[1], n_restarts_optimizer=n_restarts_optimizer)\n",
    "#gp3 = GaussianProcessRegressor(kernel=kernels[2], n_restarts_optimizer=n_restarts_optimizer)\n",
    "#gp4 = GaussianProcessRegressor(kernel=kernels[3], n_restarts_optimizer=n_restarts_optimizer)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1101,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/sklearn/gaussian_process/gpr.py:457: UserWarning: fmin_l_bfgs_b terminated abnormally with the  state: {'grad': array([ -2.88162278e+09,   3.74558408e+14,  -6.24930055e+07,\n",
      "         9.40915516e+08,  -6.24930055e+07,   1.24951289e+08,\n",
      "         1.48931257e+09,   2.43269632e+08,  -1.29105920e+07,\n",
      "        -3.25405704e+11]), 'task': b'ABNORMAL_TERMINATION_IN_LNSRCH', 'funcalls': 53, 'nit': 2, 'warnflag': 2}\n",
      "  \" state: %s\" % convergence_dict)\n"
     ]
    }
   ],
   "source": [
    "gp1.fit(X_train, y_train);\n",
    "gp2.fit(X_train, y_train);\n",
    "#gp3.fit(X, y);\n",
    "#gp4.fit(X, y);"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1102,
   "metadata": {},
   "outputs": [],
   "source": [
    "pred=np.array([list(range(0,56))]).T\n",
    "p1=gp1.predict(pred)\n",
    "p2=gp2.predict(pred)\n",
    "#p3=gp3.predict(pred)\n",
    "#p4=gp4.predict(pred)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1103,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x10c8923c8>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "###  RESULTS VARY BY RUN.  This is sampling \n",
    "plt.plot(p1 )\n",
    "#plt.plot(p2)\n",
    "#plt.plot(p3) \n",
    "#plt.plot(p4)\n",
    "#plt.scatter(X, y)\n",
    "plt.scatter(X_train, y_train)\n",
    "plt.plot(X_test, y_test,'-ok')\n",
    "mae=mean_absolute_error(y_test, p1[28:])\n",
    "mae2=mean_absolute_error(y_test, p2[28:])\n",
    "plt.title( round(mae,2))\n",
    "plt.xlim(xmin=28)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1104,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x10afad4a8>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "#gp1.kernel_\n",
    "plt.plot(X_test, y_test,'-ok')\n",
    "plt.title( round(mae2,2))\n",
    "plt.plot(p2)\n",
    "plt.xlim(xmin=28)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1105,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "249**2 * Matern(length_scale=782, nu=5) + 1.64**2 * Matern(length_scale=1, nu=5) * 1.07**2 * ExpSineSquared(length_scale=53, periodicity=7) + 3.47**2 * ExpSineSquared(length_scale=1, periodicity=7)"
      ]
     },
     "execution_count": 1105,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "gp2.kernel_"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1110,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[ 120.30874356]])"
      ]
     },
     "execution_count": 1110,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "gp1.predict(62)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "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": "3.6.3"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}