Inititial upload of files

Copula.py

@@ -0,0 +1,67 @@
+from __future__ import division
+from scipy.stats import kendalltau, pearsonr, spearmanr
+import numpy as np
+from scipy.integrate import quad
+from scipy.optmize import fmin
+from scipy.interpolate import interp1d
+import statistics as st
+
+
+class Copula():
+    """
+    This class estimates teh parameters of copula to generated joint
+    random variables for the parameters.\
+    
+    This class hsas teh following three copulas
+        Clayton
+        Frank
+        Gumbell
+    """
+
+    def __init__ (self, X, Y, family):
+
+        # check dimensions of the input arrays
+        if not ((X.nidims == 1) and (Y.ndims == 1)):
+            raise ValueError("The dimensions of array should be one")
+
+
+        # input arrays should have the same size
+        if X.size is not Y.size:
+            raise ValueError("The size of both Arrays shoudl be the same")
+
+
+        #check if the name of the copula family is correct
+        copula_family =['clayton', "frank", "gumbell"]
+        if family not in copula_family:
+            raise ValueError('The family should be clayton or frank or gumbell')
+
+        self.X = X
+        self.Y = Y
+        self.family= family
+
+        # estimate kendalls' rank correlation
+        self.tau = kendalltau(self.X, self.Y)[0]
+
+        # estimate person R and spearman R
+        self.pr = pearsonr(self.X, self.Y)[0]
+        self.pr = spearmanr(self.X, self.Y)[0]
+
+        self._get_parameters()
+
+        self.U = None
+        self.V = None
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+

copula.ipynb

@@ -0,0 +1,160 @@
+{
+ "cells": [
+  {
+   "cell_type": "code",
+   "execution_count": 3,
+   "metadata": {
+    "collapsed": false
+   },
+   "outputs": [],
+   "source": [
+    "from __future__ import division\n",
+    "from scipy.stats import kendalltau, pearsonr, spearmanr\n",
+    "import numpy as np\n",
+    "from scipy.integrate import quad\n",
+    "from scipy.optimize import fmin\n",
+    "import sys\n",
+    "#import statistics as st\n",
+    "from scipy.interpolate import interp1d\n",
+    "#from stats import scoreatpercentis"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 13,
+   "metadata": {
+    "collapsed": false
+   },
+   "outputs": [],
+   "source": [
+    "class Copula():\n",
+    "    \"\"\"\n",
+    "    This class estimats parameter of copula generated joint \n",
+    "    random variate for the parameters.\n",
+    "    This class has the following three copulas\n",
+    "        Clayton\n",
+    "        Frank\n",
+    "        Gumbell\n",
+    "    \"\"\"\n",
+    "    \n",
+    "    def __init__(self, x, y, family):\n",
+    "        \"\"\"\n",
+    "        Initialise the class with x and Y\n",
+    "        Input:\n",
+    "            X: one dimesional numpy array\n",
+    "            Y: one dimensional mumpy array\n",
+    "            family: clayton or frank or gumbell\n",
+    "            \n",
+    "            Note that the size of x and Y should be the same\n",
+    "        \"\"\"\n",
+    "        \n",
+    "        if not((x.ndim==1) and (y.ndim==1)):\n",
+    "            raise ValueError(\"The dimensions of array should \\\n",
+    "                             be one\")\n",
+    "            \n",
+    "        if x.size != y.size:\n",
+    "            raise ValueError(\"the size of both arrays \\\n",
+    "                             should be the same\")\n",
+    "            \n",
+    "        copula_family =[\"clayton\", \"frank\", \"gumbel\"]\n",
+    "        if family not in copula_family:\n",
+    "            raise ValueError(\"The family should be clayton \\\n",
+    "                             or frank of gumbell\")\n",
+    "            \n",
+    "        self.x= x\n",
+    "        self.y= y\n",
+    "        self.family = family\n",
+    "        \n",
+    "        tau = kendalltau(self.x,self.y)[0]\n",
+    "        self.tau = tau\n",
+    "        \n",
+    "        self.pr =pearsonr(self.x, self.y)[0]\n",
+    "        self.sr= spearman(self.x,self.y)[0]\n",
+    "        \n",
+    "        self._get_parameter()\n",
+    "        \n",
+    "        self.u = None\n",
+    "        self.v = None\n",
+    "          \n",
+    "    def _get_parameter(self):\n",
+    "        \"\"\"\n",
+    "        estimate the parameter (theta) of copula\n",
+    "        \"\"\"\n",
+    "        \n",
+    "        if self.family == \"clayton\":\n",
+    "            self.theta = 2*self.tau/(1-self.tau)\n",
+    "            \n",
+    "        elif self.family == \"frank\"\n",
+    "            self.theta= -fmin(self._frank_fun, -5, disp=False)[0]\n",
+    "            \n",
+    "        elif self.family == 'gumbel':\n",
+    "            self.theta = 1/(1-self.tau)\n",
+    "            \n",
+    "    def generate_uv(self.n=1000):\n",
+    "        \"\"\"\n",
+    "        generate random variables (u,v)\n",
+    "        input: n: number of random copula to be generated\n",
+    "        output: u and v: generated copula\n",
+    "        \"\"\"\n",
+    "        #Clayton copula\n",
+    "        \n",
+    "        if self.family == \"clayton\":\n",
+    "            u = np.random.uniform(size=n)\n",
+    "            w = np.random.uniform(size = n)\n",
+    "            \n",
+    "            if self.theta <= -1:\n",
+    "                raise ValueError(\"the parameter for clayton copula shoud be more than -1)\n",
+    "            if self.theta == 0:\n",
+    "                raise ValueError('The parameter for clayton copula should not be 0')\n",
+    "                             \n",
+    "            if self.theta < sys.float_info.epsilon:\n",
+    "                v=w\n",
+    "            else:\n",
+    "                v=u*(w**(-self.theta/(1+self.theta))-1+u**self.theta)**(-1/self.theta)\n",
+    "                             \n",
+    "        # Frank Copula                    \n",
+    "        elif self.family == \"frank\":\n",
+    "          u = random.uniform(size=n)\n",
+    "          w = random.uniform(size=n)\n",
+    "        \n",
+    "          if self.theta == 0:\n",
+    "              raise ValueError(\"the parameter for frank copula \\\n",
+    "              should not be 0\")\n",
+    "                             \n",
+    "          if abs(self.theta) > np.log(sys.float_info.max):\n",
+    "                v= ( u < 0) + np.sign(self.theta)*U\n",
+    "          elif abs(self.theta) > np.sqrt(sys.float_info.epsilon):\n",
+    "                                    "
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {
+    "collapsed": true
+   },
+   "outputs": [],
+   "source": []
+  }
+ ],
+ "metadata": {
+  "kernelspec": {
+   "display_name": "Python 2",
+   "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
+}