Principal Components Analysis - Revision history

Neteler: update for G7 compatibility

2014-12-28T15:00:21Z

update for G7 compatibility

← Older revision		Revision as of 15:00, 28 December 2014
Line 225:		Line 225:
	Command in one line		Command in one line

	(echo 3; r.covar -r spot.ms.1,spot.ms.2,spot.ms.3) \| m.eigensystem		(echo 3; r.covar -r spot.ms.1,spot.ms.2,spot.ms.3 \| tail -n 3) \| m.eigensystem

Neteler: update m.eigensystem

2014-12-28T14:06:18Z

update m.eigensystem

← Older revision		Revision as of 14:06, 28 December 2014
Line 105:		Line 105:
	==== Using '''''m.eigensystem''''' ====		==== Using '''''m.eigensystem''''' ====

	Note: download m.eigensystem from [[GRASS_AddOns#m.eigensystem GRASS-Addons]]		Note: download m.eigensystem from [[GRASS_AddOns#m.eigensystem GRASS-Addons]] or simply with

			g.extention m.eigensystem


	Command in one line		Command in one line

	(echo 3; r.covar spot.ms.1,spot.ms.2,spot.ms.3) \| m.eigensystem		(echo 3; r.covar spot.ms.1,spot.ms.2,spot.ms.3 \| tail -n 3) \| m.eigensystem

⚠️NikosA: /* Issues concerning the i.pca module */ Talk about fixed issues removed

2013-02-07T15:31:17Z

Issues concerning the i.pca module: Talk about fixed issues removed

← Older revision		Revision as of 15:31, 7 February 2013
Line 87:		Line 87:

	where x is a numeric or complex matrix (or data frame) which provides the data for the principal components analysis (R Documentation).		where x is a numeric or complex matrix (or data frame) which provides the data for the principal components analysis (R Documentation).

	~~== Issues concerning the ''i.pca'' module ==~~

	<strike>In general ''i.pca'' works as expected (tested with SPOT and Landsat5 satellite data from the spearfish location) producing slightly different eigenvectors. Nevertheless, the reported eigenvalues are very different than the ones reported by m.eigensystem or R's equivalent functions. In addition, using some MODIS surface reflectance bands, there is one case in which the reported eigenvectors, are not centered prior to the Principal Component Analysis as it should be done.</strike>


	~~'''Overview of points for discussion'''~~

	* <strike>Eigenvalues vary between i.pca and the rest of modules/functions</strike>

	* <strike>Data centering not performed in one case</strike> '''Fixed''' -- <span style="color:#900000">add respective ticket information!</span>

	== Useful details ==		== Useful details ==

⚠️Maurizio.m: /* Using R's princomp() function */

2012-08-17T10:19:55Z

Using R's princomp() function

← Older revision		Revision as of 10:19, 17 August 2012
Line 185:		Line 185:

	A better option is to use R's ''complete.cases()'' function. See ''?complete.cases'' within R.		A better option is to use R's ''complete.cases()'' function. See ''?complete.cases'' within R.


	Command		Command

	princomp(spot.ms.nonas)		princomp(spot.ms.nonas)


			Another possible way is to use the function na.omit() and outputs are exactly the same

			princomp(na.omit(spot.ms))

⚠️HamishBowman: /* Using i.pca */ colorize bash command

2012-06-25T22:28:48Z

Using i.pca: colorize bash command

← Older revision		Revision as of 22:28, 25 June 2012
Line 315:		Line 315:

	Command		Command
			<source lang="bash">
	i.pca input=spot.ms.1,spot.ms.2,spot.ms.3 out=pca.spot.ms rescale=0,0		i.pca input=spot.ms.1,spot.ms.2,spot.ms.3 out=pca.spot.ms rescale=0,0
			</source>

	Output		Output

⚠️NikosA: /* Issues concerning the i.pca module */

2012-04-26T22:36:04Z

Issues concerning the i.pca module

← Older revision		Revision as of 22:36, 26 April 2012
Line 97:		Line 97:
	* <strike>Eigenvalues vary between i.pca and the rest of modules/functions</strike>		* <strike>Eigenvalues vary between i.pca and the rest of modules/functions</strike>

	* <strike>Data centering not performed in one case</strike> '''Fixed''' -- add respective ticket information!		* <strike>Data centering not performed in one case</strike> '''Fixed''' -- <span style="color:#900000">add respective ticket information!</span>

	== Useful details ==		== Useful details ==

⚠️NikosA: Past issues fixed!

2012-04-26T22:32:58Z

Past issues fixed!

← Older revision		Revision as of 22:32, 26 April 2012
Line 90:		Line 90:
	== Issues concerning the ''i.pca'' module ==		== Issues concerning the ''i.pca'' module ==

	<strike>In general ''i.pca'' works as expected (tested with SPOT and Landsat5 satellite data from the spearfish location) producing slightly different eigenvectors. Nevertheless, the reported eigenvalues are very different than the ones reported by m.eigensystem or R's equivalent functions.~~</strike>~~ In addition, using some MODIS surface reflectance bands, there is one case in which the reported eigenvectors, are not centered prior to the Principal Component Analysis as it should be done. ~~'''Need to re-check if this is still an issue.'''~~		<strike>In general ''i.pca'' works as expected (tested with SPOT and Landsat5 satellite data from the spearfish location) producing slightly different eigenvectors. Nevertheless, the reported eigenvalues are very different than the ones reported by m.eigensystem or R's equivalent functions. In addition, using some MODIS surface reflectance bands, there is one case in which the reported eigenvectors, are not centered prior to the Principal Component Analysis as it should be done.</strike>


Line 97:		Line 97:
	* <strike>Eigenvalues vary between i.pca and the rest of modules/functions</strike>		* <strike>Eigenvalues vary between i.pca and the rest of modules/functions</strike>

	* Data centering not performed in one case		* <strike>Data centering not performed in one case</strike> '''Fixed''' -- add respective ticket information!

	== Useful details ==		== Useful details ==

Neteler: /* e-mails in GRASS-user mailing list */

2012-04-05T16:04:53Z

e-mails in GRASS-user mailing list

← Older revision		Revision as of 16:04, 5 April 2012
Line 694:		Line 694:

	[[Category: Documentation]]		[[Category: Documentation]]
			[[Category: Image processing]]

Neteler: urls restored

2011-11-05T06:41:36Z

urls restored

← Older revision		Revision as of 06:41, 5 November 2011
Line 684:		Line 684:
	'''Posts in grass-user mailing list'''		'''Posts in grass-user mailing list'''

	[http://n2.nabble.com/i.pca-output-td1863271.html#a1863271]		* [http://osgeo-org.1803224.n2.nabble.com/i.pca-output-td1863271.html#a1863271]
			* [http://osgeo-org.1803224.n2.nabble.com/Testing-i.pca-~-prcomp()%2C-m.eigensystem-~-princomp()-td2413700.html#a2415727 Testing i.pca ~ prcomp(), m.eigensystem ~ princomp()]
	[http://n2.nabble.com/Testing-i.pca-~-prcomp()%2C-m.eigensystem-~-princomp()-td2413700.html#a2415727 Testing i.pca ~ prcomp(), m.eigensystem ~ princomp()]		* [http://osgeo-org.1803224.n2.nabble.com/Calculating-eigen-values-and---variance-explained-after-PCA-analysis-td2383005.html#a2383165]
			* [http://osgeo-org.1803224.n2.nabble.com/Re%3A-Calculating-eigen-valuesand-varianceexplainedafter-PCA-analysis-td2413881.html#a2413881]
	[http://n2.nabble.com/Calculating-eigen-values-and---variance-explained-after-PCA-analysis-td2383005.html#a2383165]		* [http://osgeo-org.1803224.n2.nabble.com/Re%3A-Calculating-eigen-values-and--varianceexplainedafter-PCA-analysis-td2395558.html#a2409630]
			* [http://osgeo-org.1803224.n2.nabble.com/i-pca-fixes-in-trunk-td6962364.html] (2011)
	[http://n2.nabble.com/Re%3A-Calculating-eigen-valuesand-varianceexplainedafter-PCA-analysis-td2413881.html#a2413881]

	[http://n2.nabble.com/Re%3A-Calculating-eigen-values-and--varianceexplainedafter-PCA-analysis-td2395558.html#a2409630]


	'''More sources to be added'''		'''More sources to be added'''

	[[Category: Documentation]]		[[Category: Documentation]]

⚠️Gaspar33: /* Using m.eigensystem */

2011-09-26T23:56:08Z

Using m.eigensystem

← Older revision		Revision as of 23:56, 26 September 2011
Line 486:		Line 486:


	<strike>Using the <span style="color:#000090">W</span> vector, new maps can be created:</strike>. The new maps (coordinate system) system is formed by the normalized eigenvectors <span style="color:#900000">(N)</span> of the variance–covariance (or correlation) matrix.		<strike>Using the <span style="color:#000090">W</span> vector, new maps can be created:</strike>. The new maps (coordinate system) system is formed by the normalized eigenvectors <span style="color:#900000">(N)</span> of the variance–covariance (or correlation) matrix [''Tso, B. & P.M. Mather. Classification methods for remotely sensed data. 2001. Taylor & Francis, London ; New York''].

	<B>r.mapcalc</B> 'pc.1 = <span style="color:#900000">.4372107421</span>map.1 <span style="color:#900000">+.7210155161 </span>map.2 <span style="color:#900000">+ .5375717557</span>*map.3'		<B>r.mapcalc</B> 'pc.1 = <span style="color:#900000">.4372107421</span>map.1 <span style="color:#900000">+.7210155161 </span>map.2 <span style="color:#900000">+ .5375717557</span>*map.3'