Scribd
Upload
137 views3 pages

R Scripts for Data Envelopment Analysis

This document provides R code examples to illustrate data envelopment analysis (DEA). It shows how to: 1) Install an R package for DEA and load example data sets with multiple inputs and outputs; 2) Perform DEA using different assumptions like variable returns to scale, constant returns to scale, and input vs. output orientation; and 3) Calculate efficiency measures like pure technical efficiency and scale efficiency.

Uploaded by

Francisco Silva
Copyright
© All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as RTF, PDF, TXT or read online on Scribd
137 views3 pages

R Scripts for Data Envelopment Analysis

This document provides R code examples to illustrate data envelopment analysis (DEA). It shows how to: 1) Install an R package for DEA and load example data sets with multiple inputs and outputs; 2) Perform DEA using different assumptions like variable returns to scale, constant returns to scale, and input vs. output orientation; and 3) Calculate efficiency measures like pure technical efficiency and scale efficiency.

Uploaded by

Francisco Silva
Copyright
© All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as RTF, PDF, TXT or read online on Scribd

## R code to illustrate data envelopment analysis

## Install a DEA package. When asked to select a CRAN mirror,


## scroll down to find a nearby mirror, such as USA (MA).
[Link]("nonparaeff")
## Load the package
library(nonparaeff)
## Example 1: one output and one input
## Load data on DMUs' outputs and inputs. Putting the command in
parentheses
## causes the object created to be displayed.
(YXtable = [Link]("[Link]"))
## Make the variables in YXtable accessible to R
attach(YXtable)
## Plot the data
plot(x,y)
## prepare data for use by nonparaeff (outputs listed before inputs)
[Link] = [Link](y, x)
## Assuming a free disposal hull (fdh), implement an input-oriented analysis
## (orientation 1).
(fdhio11 = fdh([Link], noutput=1, orientation=1))
## Still assuming a free disposal hull, implement an output-oriented analysis
## (orientation 2).
(fdhoo11 = fdh([Link], noutput=1, orientation=2))
## With the output-orientation, nonparaeff calculates the factor by which
## a DMU could increase its output without needing additional inputs. The
## reciprocal of this factor is output-oriented technical efficiency.
1/fdhoo11

## Assuming a convex hull with variable returns to scale (rts=2), implement


an
## input-oriented analysis.
(vrio11 = dea([Link], noutput=1, orientation=1, rts=2))
## In table printed by the preceding command, eff is pure input-oriented
## technical efficiency. (It is pure in the sense of measuring efficiency
## conditional on the scale of production--that is, ignoring any
## inefficiency due to scale.
## Still assuming variable returns to scale, do an output-oriented analysis.
(vroo11 = dea([Link], noutput=1, orientation=2, rts=2))
1/vroo11$eff
## Assuming constant returns to scale (rts=1), do an input-oriented analysis.
(crio11 = dea([Link], noutput=1, orientation=1, rts=1))
## Maintaining the CRS assumption, do an output-oriented analysis.
(croo11 = dea([Link], noutput=1, orientation=2, rts=1))
1/croo11$eff
## Calculate input-oriented scale efficiency.
(seio11 = crio11$eff/vrio11$eff)
## Calculate output-oriented scale efficiency.
(seoo11 = (1/croo11$eff)/(1/vroo11$eff))

## Example 2: one output and two inputs


## Load data on DMUs' outputs and inputs.
(QKLtable = [Link]("[Link]"))
## Make the variables in YXtable accessible to R
attach(QKLtable)
## prepare data for use by nonparaeff (outputs listed before inputs)

[Link] = [Link](Q, K, L)
## ## Assuming a convex hull with variable returns to scale, implement an
## input-oriented analysis.
(vrio12 = dea([Link], noutput=1, orientation=1, rts=2))

## Example 3: two outputs and two inputs


## Load data on DMUs' outputs and inputs
(YZKLtable = [Link]("[Link]"))
attach(YZKLtable)
## prepare data for use by nonparaeff (outputs listed before inputs)
[Link] = [Link](Y, Z, K, L)
## ## Assuming a convex hull with constant returns to scale, implement an
## input-oriented analysis.
(crio22 = dea([Link], noutput=2, orientation=1, rts=1))

You might also like