jcrubino

# JSON IMDbPY class wrapper
# Goal:  wraps IMDB class data function to provide fully complaint json
# author: james , rubino <at> gmail , com
# note: contains wrapper and randomized test cases

from random import randint as r
import json
import imdb
ia = imdb.IMDb()

jcrubino

# SAPD Neighborhood Watch Db
# Parses San Antonio Police Neighborhood Calls pdfs into json for dbs
#
#
# Dev Env:  Ubuntu 12.04
# Licensed under the GNU General Public License: http://www.gnu.org/licenses/gpl.html
# Requires  xpdf utils available at the command line (not a python library (yet))
#           pymongo & Mongodb
#           pdfs from http://www.sanantonio.gov/neighborhoodpolicecalls/policecalls.aspx
#

jcrubino

This is a project to provide police event mapping and statistics for the citizens of San Antonio.
This project is meant to provide citizens with data to enable them to better watch their neighborhoods.
It was created after reading national best places to live rankings for cities.  San Antonio was hindered by a nationally high ranking property theft rate.

Project Name Ideas: 

   * Nicolas
     * Means "Victory of the People"
     * Refers to: St Nicolas of Myra; protector from theft and more generally protector of the innocent

jcrubino

I got the crazy idea that the project could be released as a Puppy / DSL linux or #! Linux distro
  A debian or ubuntu compatible distro would be ideal, a tiny linux is very seductive.

    1) Code Updates could be pushed to the nodes
    2) A compute engine can be installed and used for grid operations (There goes the dream of tiny)
       MOsix, OpenMosix, Rocks are all compute engines
       SaltStack might be ideal, hacking a new setup with zeromq might be worth the effort.
    3) Volunteers can run the distro in a vm maintaining their own computers and volunteering compute time.

I spent today looking into GIS shapefile utils

jcrubino

.. highlightlang:: rest

Welcome to Crime Crunchers!

Time to Crunch Crime!

jcrubino

# coding: utf-8
 
"""Online learning."""

import numpy as np
from numpy import sign
import itertools as it
from numpy import array as A, zeros as Z
import math

jcrubino

"""
Code for training RBMs with contrastive divergence. Tries to be as
quick and memory-efficient as possible while utilizing only pure Python
and NumPy.
"""

# Copyright (c) 2009, David Warde-Farley
# All rights reserved.
#
# Redistribution and use in source and binary forms, with or without

jcrubino

General

Tutorials on topics in machine learning

Hashing Searching Sketching.

Parallel Machine Learning

http://graphlab.org/

AdaBoost

jcrubino

class OnlineLearner(object):
    def __init__(self, **kwargs):
        self.last_misses = 0.
        self.iratio = 0.
        self.it = 1.
        self.l = kwargs["l"]
        self.max_ratio = -np.inf
        self.threshold = 500.

    def hinge_loss(self, vector, cls, weight):

jcrubino

#!/usr/bin/env python
# -*- coding: utf-8 -*-

from __future__ import with_statement
import collections, operator, math, random, pprint

class Classifier(object):
    AttrsToDump = ["value_counts", "class_counts", "features", "feature_counts"]

    def __init__(self, features={}, verbose=False):