Closures for Data Science

closures_for_data_science.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Closures for Data Science\n",
    "------"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Problem\n",
    "-----\n",
    "\n",
    "Sometimes you have data that is outside of valid bounds and has to be trimmed. If the data below a value, it should be changed to take the minimum value. If the data is above a value, it should be change to take on the maximum value.\n",
    "\n",
    "For example, RBG values can only be from 0 to 255:\n",
    "\n",
    "![](http://tutorial.techaltum.com/images/css-colors.jpg)\n",
    "\n",
    "Here is a bound function:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "def bound(value, min_value, max_value):\n",
    "    \"Limit value between the min and max\"\n",
    "    return min(max_value, max(value, min_value))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [],
   "source": [
    "assert bound(42, min_value=0, max_value=255) == 42\n",
    "assert bound(-1, min_value=0, max_value=255) == 0\n",
    "assert bound(256, min_value=0, max_value=255) == 255"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Python Function Superpower:  <br> A Python function can return any object, including an another function!\n",
    "--------\n",
    "\n",
    "Let's use that feature to create a variation of our bound function"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "def make_bound_function(min_value, max_value):\n",
    "    \"Define a bound function with a certain min and max\"\n",
    "    def bound(value):\n",
    "        \"Limit value between the min and max\"\n",
    "        return min(max_value, max(value, min_value))\n",
    "    return bound"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Take moment to wrap your head around the idea that the outer function `make_bound_func` creates and returns the inner function `bound`"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# Call the outer function to define/return the inner fuction\n",
    "bound_rbg = make_bound_function(min_value=0, max_value=255)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "function"
      ]
     },
     "execution_count": 32,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# The outer function returns the inner function\n",
    "type(bound_rbg)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "42\n",
      "0\n",
      "255\n"
     ]
    }
   ],
   "source": [
    "# Call the inner function to bound the values\n",
    "print(bound_rbg(42))\n",
    "print(bound_rbg(-1))\n",
    "print(bound_rbg(256))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "This is a closure, a function wrapped in another function.\n",
    "\n",
    "The wrapping __binds__ the values/state to the inner function."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Another Problem\n",
    "----\n",
    "\n",
    "Let's say our system can only take non-negative values.\n",
    "\n",
    "We can define another bound function that forces all negative numbers to be zero."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# Let's call the outer function again to define/return a different inner fuction\n",
    "from math import inf\n",
    "\n",
    "bound_non_negative = make_bound_function(min_value=0, max_value=inf)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "42\n",
      "0\n",
      "1000000\n"
     ]
    }
   ],
   "source": [
    "# Call the inner function to bound the values\n",
    "print(bound_non_negative(42))\n",
    "print(bound_non_negative(-1))\n",
    "print(bound_non_negative(1_000_000))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Summary\n",
    "-----\n",
    "\n",
    "Closures are not required in Python. You can just pass a series of arguments to a function.\n",
    "\n",
    "Closures are a nice-to-have tool that can simplify complex function calls and create guard-rails for programming. Closures help build large, concurrent systems by \"freezing\" functions and limiting the dynamic nature of Python."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<br>\n",
    "<br> \n",
    "<br>\n",
    "\n",
    "----"
   ]
  }
 ],
 "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": "3.6.1"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}