Prabhas Pokharel prabhasp

## index.Rmd
We'll be checking whether the top 100 users of formhub follow a power law distribution, or a log-normal distribution, using guidance from [a blog post by CMU stats prof. Cosma Rohilla Shaliz](http://vserver1.cscs.lsa.umich.edu/~crshalizi/weblog/857.html).

First, load up the dataset; I have included it here in case you want to re-produce it. Each value is a number of submissions on formhub for top 100 users.

```{r}
source("~/Downloads/pli-R-v0.0.3-2007-07-25/pareto.R")
source("~/Downloads/pli-R-v0.0.3-2007-07-25/lnorm.R")
source("~/Downloads/pli-R-v0.0.3-2007-07-25/power-law-test.R")

users <- c(220346L, 31099L, 28568L, 16573L, 14862L, 7531L, 6510L, 6138L,

## unnamed-chunk-1.png

      
              10 files
            
          
              0 forks
            
          
              0 comments
            
          
              0 stars
            
          
                prabhasp
                / unnamed-chunk-1.png
            
            
              Last active
              December 11, 2015 02:19
            
              
                Coverage analysis for Nigeria -- generated by Rmarkdown.
              
          
## KTMJS.md

      
              15 files
            
          
              0 forks
            
          
              0 comments
            
          
              0 stars
            
          
                prabhasp
                / KTMJS.md
            
            
              Last active
              August 29, 2015 14:18
            
          
    KTMJS Talk by @prabhasp

  
## index.html
<!DOCTYPE html>
<html>
<head>
<title>Foo</title>
<meta charset='utf-8' />
<meta name='viewport' content='width=device-width, initial-scale=1.0, maximum-scale=1.0, user-scalable=0' />
<style type='text/css'>
body {
  background:#000;
  color:#fff;

## yatayat_flipped.json

      
              1 file
            
          
              0 forks
            
          
              0 comments
            
          
              0 stars
            
          
                prabhasp
                / yatayat_flipped.json
            
            
              Created
              March 5, 2015 14:57
            
          
      Sorry, something went wrong. Reload?
      Sorry, we cannot display this file.
      Sorry, this file is invalid so it cannot be displayed.
      
          Viewer requires iframe.
      
    
## index.Rmd
<link href="http://kevinburke.bitbucket.org/markdowncss/markdown.css" rel="stylesheet">

Finding indicator dependencies
========================================================
I had been meaning to look into R's metaprogramming features for a while now. I finally had a chance today (thanks [Hadley](http://adv-r.had.co.nz/)!), and I used it to experiment towards a problem that had been in the back of my mind for a while: finding depencies within indicator definitions.

Below, I implement a find dependencies function, which takes a set of indicators, and finds dependencies within it. Indicators are fields within a dataset, some of which are already there, and some of which are newly created. The dependency finding problem is investigating which new indicators derive from which existing ones. We think of these relationships as dependencies: for an indicator such as pupil-to-teacher-ratio (defined as the number-of-pupils divided by the number-of-teachers), pupil-to-teacher-ratio is dependent on number-of-pupils

## PlotDKs.R
get_dk_reason <- function(df) {
    dk <- names(df)[str_detect(names(df), "dont")]
    llply(df[dk], function(x) { as.character(na.exclude(x)) })
}
plot_percent_dks <- function(dk_list, N) {
    # d will be a list of question name and length
    d <- ldply(dk_list, length)
    # order the data frame
    d <- arrange(d, V1)
    # divide by N (which is supposed to be total responses
	We'll be checking whether the top 100 users of formhub follow a power law distribution, or a log-normal distribution, using guidance from [a blog post by CMU stats prof. Cosma Rohilla Shaliz](http://vserver1.cscs.lsa.umich.edu/~crshalizi/weblog/857.html).

	First, load up the dataset; I have included it here in case you want to re-produce it. Each value is a number of submissions on formhub for top 100 users.

	```{r}
	source("~/Downloads/pli-R-v0.0.3-2007-07-25/pareto.R")
	source("~/Downloads/pli-R-v0.0.3-2007-07-25/lnorm.R")
	source("~/Downloads/pli-R-v0.0.3-2007-07-25/power-law-test.R")

	users <- c(220346L, 31099L, 28568L, 16573L, 14862L, 7531L, 6510L, 6138L,
	<!DOCTYPE html>
	<html>
	<head>
	<title>Foo</title>
	<meta charset='utf-8' />
	<meta name='viewport' content='width=device-width, initial-scale=1.0, maximum-scale=1.0, user-scalable=0' />
	<style type='text/css'>
	body {
	background:#000;
	color:#fff;
	<link href="http://kevinburke.bitbucket.org/markdowncss/markdown.css" rel="stylesheet">

	Finding indicator dependencies
	========================================================
	I had been meaning to look into R's metaprogramming features for a while now. I finally had a chance today (thanks [Hadley](http://adv-r.had.co.nz/)!), and I used it to experiment towards a problem that had been in the back of my mind for a while: finding depencies within indicator definitions.

	Below, I implement a find dependencies function, which takes a set of indicators, and finds dependencies within it. Indicators are fields within a dataset, some of which are already there, and some of which are newly created. The dependency finding problem is investigating which new indicators derive from which existing ones. We think of these relationships as dependencies: for an indicator such as pupil-to-teacher-ratio (defined as the number-of-pupils divided by the number-of-teachers), pupil-to-teacher-ratio is dependent on number-of-pupils
	get_dk_reason <- function(df) {
	dk <- names(df)[str_detect(names(df), "dont")]
	llply(df[dk], function(x) { as.character(na.exclude(x)) })
	}
	plot_percent_dks <- function(dk_list, N) {
	# d will be a list of question name and length
	d <- ldply(dk_list, length)
	# order the data frame
	d <- arrange(d, V1)
	# divide by N (which is supposed to be total responses