datafitting.html

<html>

<head>
<meta HTTP-EQUIV="Content-Type" CONTENT="text/html; charset=utf-8">
   <meta NAME="Description" CONTENT="Iterative Proportional Fitting">
   <meta NAME="AUTHOR" CONTENT="Iterative Proportional Fitting">
   <meta NAME="KEYWORDS" CONTENT="Iterative Proportional Fitting, raking">
<title>Iterative Proportional Fitting</title>
<link rel="shortcut icon" href="https://github.com/edyhsgr/edyhsgr.github.io/raw/master/images/Icon.png">
   
<script>
  (function(i,s,o,g,r,a,m){i['GoogleAnalyticsObject']=r;i[r]=i[r]||function(){
  (i[r].q=i[r].q||[]).push(arguments)},i[r].l=1*new Date();a=s.createElement(o),
  m=s.getElementsByTagName(o)[0];a.async=1;a.src=g;m.parentNode.insertBefore(a,m)
  })(window,document,'script','https://www.google-analytics.com/analytics.js','ga');

  ga('create', 'UA-9665883-1', 'auto');
  ga('send', 'pageview');
</script>

</head>

<body TEXT="#505050"  
 LINK="006699" ALINK="006699" VLINK="006699">

<table WIDTH="100%" NOSAVE >
<table BORDER=0 CELLPADDING=0 align="center" valign="bottom">

<tr NOSAVE>
<td NOSAVE>
<div align=center> 
<br>

<table WIDTH="100%" NOSAVE >
<table BORDER=0 CELLPADDING=0 align="center" valign="bottom">
<blockquote>
<td valign=top width=750 >
<font FACE="HELVETICA">
<font size=3">

<br>
 <br>

<u><b>Iterative Proportional Fitting Information, Code, and Links</b></u>
</br><span style="display:block; margin-bottom:-15px;"></span></br>

<strong><a STYLE="text-decoration:none" 
href="https://edyhsgr.github.io/IPFDescription/AKDOLWDIPFTWOD.pdf" target="_blank">Iterative Proportional Fitting for a Two-Dimensional Table</a>
</strong><br>
An informal description from May 2008 of two-dimensional IPF that I 
(<a STYLE="text-decoration:none" href="https://edyhsgr.github.io/">Eddie Hunsinger</a>) 
prepared for my coworkers and fellow researchers. Please feel free to offer comments or corrections.
IPF is a big name for a pretty simple (but terrific) technique-- it is used quite a bit, but I think it could be used more. <a STYLE="text-decoration:none" 
href="http://www.google.com/search?q=nels+tomlinson" target="_blank">Nels Tomlinson</a> handed down much of any knowledge and computer code that I have on IPF.
<br><br>

<strong><a STYLE="text-decoration:none" 
href="https://edyhsgr.github.io/IPFDescription/AKDOLWDIPFTHREED.pdf" target="_blank">Iterative Proportional Fitting for a Three-Dimensional Table</a>
</strong><br>
A description of three-dimensional IPF (May 2008).
<br><br>

<strong><a STYLE="text-decoration:none" 
href="https://edyhsgr.github.io/IPFDescription/AKDOLWDIPFFOURD.pdf" target="_blank">Iterative Proportional Fitting for a Four-Dimensional Table</a>
</strong><br>
A description of four-dimensional IPF (May 2008).
<br><br>

<strong><a STYLE="text-decoration:none" 
href="https://edyhsgr.github.io/IPFDescription/IPF.zip">Iterative Proportional Fitting R Code</a>
</strong><br>
<a STYLE="text-decoration:none;display:inline;padding:0" href="http://www.r-project.org/" target="_blank">R</a>
code, supporting files and instruction/documentation to perform two-, three- and four-dimensional IPF. 
Though I did pieces of the development for the IPF functions, the heavy lifting (the bulk of the 2D and 3D functions) was done by 
<a STYLE="text-decoration:none;display:inline;padding:0" href="http://www.google.com/search?q=nels+tomlinson" target="_blank">Nels Tomlinson</a>, 
my predecessor at the 
<a STYLE="text-decoration:none;display:inline;padding:0" href="http://labor.state.ak.us/" target="_blank">Alaska Department of Labor and Workforce Development</a>.
<br><br>
                                                                          
<strong><a STYLE="text-decoration:none" 
href="https://github.com/AppliedDemogToolbox/Sprague_IPFSAS/raw/master/ipf2way.sas" target="_blank">Webb Sprague's Iterative Proportional Fitting SAS Code</a>
</strong><br>
A great 
<a STYLE="text-decoration:none" href="http://www.sas.com/" target="_blank">SAS</a>
macro for two-dimensional iterative proportional fitting, developed by Webb Sprague, a researcher with the State of Washington. 
Please note this code has only been tested with some sample data (August 2011).
<br><br>
                                                                          
<strong><a STYLE="text-decoration:none" 
href="https://github.com/Dirguis/ipfn" target="_blank">Damien Forthommme's ipfn: Iterative Proportional Fitting With N Dimensions For Python</a>
</strong><br>
From the README, "Iterative proportional fitting is an algorithm used is many different fields such as economics or social sciences, 
to alter results in such a way that aggregates along one or several dimensions match known marginals (or aggregates along these same dimensions)." 
The package includes
<a STYLE="text-decoration:none;display:inline;padding:0" href="https://numpy.org/" target="_blank">NumPy</a> 
and 
<a STYLE="text-decoration:none;display:inline;padding:0" href="https://pandas.pydata.org/" target="_blank">pandas</a> 
versions of the algorithm. 
<a STYLE="text-decoration:none;display:inline;padding:0" href="https://github.com/edyhsgr/IPFCompare/blob/main/ipfnCompare.ipynb" target="_blank">Here</a> 
is a link to application and comparison of it in a 
<a STYLE="text-decoration:none;display:inline;padding:0" href="https://jupyter.org/" target="_blank">Jupyter Notebook</a> (January 2023).
<br><br>
                                                                          
<strong><a STYLE="text-decoration:none" 
href="https://edyhsgr.github.io/documents/GQControlExample.pdf" target="_blank">A Particular IPF Application</a>
</strong><br>
An April 2011 message from me with description on use of IPF to reasonably fit 
sampled "Group Quarters" (GQ) (shared living facilities, as opposed to traditional households) data to 
separately-estimated GQ-type (e.g. prisons, dorms, etc.), 
GQ-county (location) and GQ-age (age of resident) control/marginal totals, given sparse and tightly-clustered sample-cell-data by type and county.
<br><br>
                        
<strong><a STYLE="text-decoration:none" 
href="http://www.stat.auckland.ac.nz/~lee/SNZ.pdf" target="_blank">Alan Lee's Generating Synthetic Unit-Record Data from Published Marginal Tables</a>
</strong><br>
A May 2007 paper by Alan Lee, a professor from the University of Auckland, that covers IPF and other techniques for cell estimation based on marginal totals, and offers computer code to perform these techniques. 
<br><br>

<strong><a STYLE="text-decoration:none" 
href="http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.17.2073" target="_blank">Paul Norman's Putting Iterative Proportional Fitting on the Researcher's Desk</a>
</strong><br>
An October 1999 paper by Paul Norman, a professor at the University of Leeds, that describes IPF and an MS Excel macro to perform IPF on two-dimensional tables.
<br><br>

<strong><a STYLE="text-decoration:none" 
href="https://projecteuclid.org/journals/annals-of-mathematical-statistics/volume-11/issue-4/On-a-Least-Squares-Adjustment-of-a-Sampled-Frequency-Table/10.1214/aoms/1177731829.full" target="_blank">Adjustment of a Sampled Frequency Table When the Expected Marginal Totals are Known</a>
</strong><br>
A useful and historically important text about IPF from December 1940 - pp 439-442 presents an early documentation of the technique. 
Some errors related to interpretation in it have been identified since its publication. And 
<a STYLE="text-decoration:none;display:inline;padding:0" href="https://www.delpher.nl/nl/tijdschriften/view?coll=dts&identifier=dts:2980078:mpeg21" target="_blank">here</a> 
is an earlier (1937) documentation of the technique by a researcher named Kruithoff. 

<br><br><br>

<a STYLE="text-decoration:none" href="https://edyhsgr.github.io">Eddie Hunsinger</a>
<br><br>
</div>

<br><br>

</body>

</html>