What if we could use [the amazing] Clusterpy on top of Pysal?

clusterpysal.ipynb

no iterations.ipynb

{
 "metadata": {
  "name": "",
  "signature": "sha256:0333f9a9480cfb8b2972406747e3000ab4ff142852f2ee25a6cfd404e95953e3"
 },
 "nbformat": 3,
 "nbformat_minor": 0,
 "worksheets": [
  {
   "cells": [
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "import numpy as np\n",
      "import pysal as ps"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "(ImportError('cannot import name minimize_scalar',), 'Maximum Likelihood in PySAL requires SciPy version 0.11 or newer.')\n",
        "(ImportError('cannot import name minimize_scalar',), 'Maximum Likelihood in PySAL requires SciPy version 0.11 or newer.')\n"
       ]
      }
     ],
     "prompt_number": 1
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "class Utils(object):\n",
      "    \"\"\"\n",
      "    \"\"\"\n",
      "    @staticmethod\n",
      "    def numeric_dbf(dbf, attributes = None):\n",
      "        \"\"\"\n",
      "        \"\"\"\n",
      "        wanted_attr = None\n",
      "        if attributes:\n",
      "            wanted_attr = np.in1d(dbf.header, attributes)\n",
      "        else:\n",
      "            wanted_attr = ~np.in1d(dbf.header, ['ID', 'Id', 'id' ])\n",
      "        \n",
      "        num_types = np.in1d([ i[0] for i in dbf.field_spec], ['N', 'F'])\n",
      "        attr_indices = np.nonzero(num_types & wanted_attr)\n",
      "\n",
      "        num_fields = len(attr_indices[0])\n",
      "        mat = np.empty((dbf.n_records, num_fields))\n",
      "\n",
      "        for row in range(dbf.n_records):\n",
      "            row_attrs = np.array(dbf.by_row(row))\n",
      "            mat[row,:] = row_attrs[attr_indices].astype(np.float)\n",
      "        return mat\n",
      "\n",
      "class cluster_execution(object):\n",
      "    \"\"\"\n",
      "    \"\"\"\n",
      "    def __init__(self, w, dbf):\n",
      "        \"\"\"\n",
      "        \"\"\"\n",
      "        self.w = w\n",
      "        self.dbf = dbf\n",
      "        self.attr_matrix = Utils.numeric_dbf(dbf)\n",
      "        self.regions = np.array([-1] * dbf.n_records, dtype=numpy.short)\n",
      "    \n",
      "    def region_centroid(self, region):\n",
      "        \"\"\"\n",
      "        \"\"\"\n",
      "        areas = np.nonzero(self.regions == region)\n",
      "        region = np.array(self.attr_matrix[areas])\n",
      "        return region.sum(axis=0) / len(areas)\n",
      "    \n",
      "    def closest_to_region(self, areas, region):\n",
      "        \"\"\"\n",
      "        \"\"\"\n",
      "        distance = np.vectorize(self.distance_area_to_region, otypes=[np.ndarray])\n",
      "        # convert to type object to be able to send an array as parameter to a vectorized function\n",
      "        centroid_obj = np.array((1,), dtype=object)\n",
      "        centroid_obj[0] = self.region_centroid(region)\n",
      "        closest = distance(areas, region, centroid_obj).argmin()\n",
      "        return areas[closest]\n",
      "\n",
      "    def distance_area_to_region(self, area, region, centroid = None):\n",
      "        \"\"\"\n",
      "        Difference squared\n",
      "        \"\"\"\n",
      "        if centroid == None:\n",
      "            areas = np.nonzero(self.regions == region) #area in region?\n",
      "            centroid = self.region_centroid(region)\n",
      "        diff = self.attr_matrix[area] - centroid\n",
      "        return np.sum(diff ** 2)"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 2
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "data = 'china'\n",
      "data = 'n100'\n",
      "shpf = data + '.shp'\n",
      "dbff = data + '.dbf'"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 3
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "dbf = ps.open(dbff)\n",
      "shp = ps.open(shpf)"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 4
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "### Create a cluster execution with a W and DBF"
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "cl = cluster_execution(ps.rook_from_shapefile(shpf), dbf)"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 5
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "cl.regions[:2] = 0\n",
      "cl.regions"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "metadata": {},
       "output_type": "pyout",
       "prompt_number": 6,
       "text": [
        "array([ 0,  0, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1,\n",
        "       -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1,\n",
        "       -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1,\n",
        "       -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1,\n",
        "       -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1,\n",
        "       -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1], dtype=int16)"
       ]
      }
     ],
     "prompt_number": 6
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "cl.region_centroid(0)"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "metadata": {},
       "output_type": "pyout",
       "prompt_number": 7,
       "text": [
        "array([ -1.6274063,  26.2722547])"
       ]
      }
     ],
     "prompt_number": 7
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "cl.distance_area_to_region(13, 0)"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "metadata": {},
       "output_type": "pyout",
       "prompt_number": 8,
       "text": [
        "155.96353719184179"
       ]
      }
     ],
     "prompt_number": 8
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "cl.closest_to_region([20, 21, 22, 23], 0)"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "metadata": {},
       "output_type": "pyout",
       "prompt_number": 9,
       "text": [
        "22"
       ]
      }
     ],
     "prompt_number": 9
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 9
    }
   ],
   "metadata": {}
  }
 ]
}