Closed mboussaa closed 7 years ago
You mean change the ellipse's confidence level ? That's not possible for the moment.
Would it be something useful for you ?
Hi @juba, I was not may be clear. I was wondering if it was possible to draw the ellipse's confidence level in my PCA individuals scrore plot... Actually, there is no option I can enable to show it
Ellipses in explor are only used when you color your points according to their type (active / supplementary) or by the levels of a qualitative supplementary variable. Ellipses are then optionally used to show the points repartition for each level.
So the option to show ellipses will only appear if you have supplementary individuals or supplementary qualitative variables in your PCA.
What I would do using explor is to to draw one confidence ellipse to detect the outliers (individuals outside the ellipse with very high variation) I don't have qualitative or supplementary variables for instance. Do you have any suggestion @juba ? Thank you
I would create for example 5 groups. Each group corresponds to a subset of my data with a label. I was thinking about using supplimentary qualitative variables to draw these 5 groups + corresponding ellipses. Is it possible?
Yes, absolutely. Add your group variable as supplementary in your PCA, and you should be able to draw ellipses in explor
. As for a way to detect outliers using confidence ellipses, I don't really know a way to do this.
Hi @juba, I added my group variable as supplementary in my PCA, yet explor won't show the groups in the individuals graph, in can't put colors or ellipses :
The groups appear in the variables graph as dots though :
Can you please help me on this issue ?
Hmm that's strange. Did you try with the development version ? And if it still doesn't work, could you share the code you use to compute your PCA ?
I just tried with the development version and it still does not work. You will find enclosed the code i use to compute my PCA. There is not much, i do just use the PCA function of the FactoMineR package. I don't know if it is really usefull but i also joined the data table used for the PCA.
> don
BAB BAC BAF BAO BAP BBA OP
CF1 0.000000798 0.0000000000 0.0000020000 0.0000000000 0.0000000000 0.0000003990 1
CF2 0.000000562 0.0000000000 0.0000000000 0.0000000000 0.0000000000 0.0000000000 2
CF3 0.000002750 0.0000002120 0.0000046600 0.0000000000 0.0000000000 0.0000004230 3
CF6 0.000000883 0.0000000000 0.0000000981 0.0000000000 0.0000000000 0.0000000981 6
CC1 0.000012100 0.0000016200 0.0000138000 0.0000015400 0.0000006840 0.0000020500 1
CC2 0.000005750 0.0000003030 0.0000194000 0.0000027300 0.0000021200 0.0000093900 2
CC5 0.000020100 0.0000000000 0.0000240000 0.0000008720 0.0000008720 0.0000065400 5
CC6 0.000008380 0.0000017100 0.0000202000 0.0000012000 0.0000000000 0.0000063300 6
CC8 0.000011900 0.0000000000 0.0000195000 0.0000000000 0.0000000000 0.0000008470 8
CC9 0.000009340 0.0000027900 0.0000055800 0.0000005370 0.0000003220 0.0000016100 9
Cor1 0.000002680 0.0000000925 0.0000057300 0.0000001850 0.0000000000 0.0000024000 1
Cor2 0.000003410 0.0000004350 0.0000070000 0.0000005070 0.0000006890 0.0000022100 2
Cor6 0.000000669 0.0000000502 0.0000018200 0.0000000251 0.0000000251 0.0000003180 6
Cor8 0.000007410 0.0000000000 0.0000094100 0.0000006660 0.0000007490 0.0000019100 8
CMO2 0.000014500 0.0000000000 0.0000205000 0.0000000000 0.0000003160 0.0000037900 2
CMO3 0.000004860 0.0000008090 0.0000146000 0.0000000000 0.0000000000 0.0000154000 3
CMO6 0.000007960 0.0000007070 0.0000087600 0.0000000000 0.0000000000 0.0000011300 6
Dar1 0.000008110 0.0000005070 0.0000120000 0.0000003380 0.0000001690 0.0000010100 1
Dar2 0.000006800 0.0000000000 0.0000109000 0.0000004950 0.0000004950 0.0000136000 2
Dar5 0.000011400 0.0000005160 0.0000114000 0.0000015500 0.0000005160 0.0000098000 5
Dar8 0.000014200 0.0000000000 0.0000283000 0.0000037800 0.0000018900 0.0000018900 8
DC1 0.000005660 0.0000008710 0.0000080500 0.0000015200 0.0000013100 0.0000100000 1
DC2 0.000004800 0.0000006850 0.0000305000 0.0000006850 0.0000010300 0.0000110000 2
DC6 0.000007010 0.0000006680 0.0000180000 0.0000000000 0.0000000000 0.0000020000 6
DC8 0.000004910 0.0000000000 0.0000038200 0.0000000000 0.0000008190 0.0000117000 8
Fra2 0.000000000 0.0000000000 0.0000123000 0.0000000000 0.0000000000 0.0000129000 2
Fra6 0.000005630 0.0000005970 0.0000085300 0.0000001710 0.0000000000 0.0000017100 6
Fra8 0.000004600 0.0000007670 0.0000222000 0.0000007670 0.0000007670 0.0000084400 8
Gen2 0.000006100 0.0000000000 0.0000033300 0.0000000000 0.0000002770 0.0000111000 2
Gen6 0.000005740 0.0000004100 0.0000344000 0.0000000000 0.0000000000 0.0000024600 6
Gri1 0.000014800 0.0000015400 0.0000300000 0.0000025700 0.0000018900 0.0000030900 1
Gri2 0.000002240 0.0000002430 0.0000081300 0.0000001820 0.0000002430 0.0000012700 2
Gri5 0.000012600 0.0000015800 0.0000174000 0.0000015800 0.0000015800 0.0000182000 5
Gri6 0.000001730 0.0000000230 0.0000024600 0.0000001150 0.0000000461 0.0000017500 6
Jon1 0.000009420 0.0000015400 0.0000084600 0.0000005770 0.0000005770 0.0000034600 1
Jon2 0.000004090 0.0000008180 0.0000028600 0.0000000000 0.0000000000 0.0000057300 2
Jon4 0.000000988 0.0000000000 0.0000050800 0.0000000000 0.0000000000 0.0000039500 4
Jon6 0.000002160 0.0000007190 0.0000024000 0.0000000000 0.0000000000 0.0000014400 6
L(3)2 0.000036100 0.0000024600 0.0000312000 0.0000016400 0.0000016400 0.0000082100 2
L(3)6 0.000009420 0.0000015700 0.0000317000 0.0000009420 0.0000000000 0.0000025100 6
L(3)9 0.000013300 0.0000028400 0.0000094800 0.0000000000 0.0000000000 0.0000000000 9
L(9)1 0.000012100 0.0000038200 0.0000124000 0.0000009000 0.0000004500 0.0000072000 1
L(9)2 0.000005350 0.0000005040 0.0000067600 0.0000003030 0.0000001010 0.0000006050 2
L(9)6 0.000003790 0.0000001110 0.0000019100 0.0000000000 0.0000000000 0.0000014400 6
Mey1 0.000003800 0.0000004950 0.0000031000 0.0000001870 0.0000000881 0.0000007050 1
Mey2 0.000000586 0.0000000451 0.0000033300 0.0000000000 0.0000000000 0.0000009010 2
Mey3 0.000003630 0.0000008060 0.0000076500 0.0000000000 0.0000004030 0.0000088600 3
Mey6 0.000002880 0.0000007200 0.0000101000 0.0000001440 0.0000000000 0.0000014400 6
Mey7 0.000008710 0.0000022700 0.0000071900 0.0000000000 0.0000000000 0.0000071900 7
Mey8 0.000001510 0.0000004230 0.0000054600 0.0000005650 0.0000003290 0.0000019800 8
Mio1 0.000012400 0.0000012000 0.0000047900 0.0000000000 0.0000000000 0.0000036000 1
Mio2 0.000000372 0.0000000000 0.0000208000 0.0000000000 0.0000000000 0.0000000000 2
NS1 0.000008050 0.0000029300 0.0000227000 0.0000007320 0.0000000000 0.0000000000 1
NS2 0.000001040 0.0000006060 0.0000078000 0.0000000000 0.0000000000 0.0000002600 2
NS6 0.000001270 0.0000004770 0.0000103000 0.0000000000 0.0000000000 0.0000001590 6
Oul1 0.000010200 0.0000051200 0.0000200000 0.0000004260 0.0000008530 0.0000012800 1
Oul2 0.000007450 0.0000005730 0.0000207000 0.0000008020 0.0000006870 0.0000012600 2
Oul6 0.000004520 0.0000000000 0.0000155000 0.0000004720 0.0000004720 0.0000002020 6
RP1 0.000014900 0.0000011900 0.0000128000 0.0000006600 0.0000002640 0.0000052800 1
RP2 0.000011300 0.0000017600 0.0000159000 0.0000002520 0.0000000000 0.0000083100 2
RP3 0.000006090 0.0000000000 0.0000094000 0.0000000000 0.0000000000 0.0000027700 3
RP6 0.000008350 0.0000096000 0.0000284000 0.0000004170 0.0000000000 0.0000075100 6
RP7 0.000025300 0.0000055700 0.0000497000 0.0000000000 0.0000006430 0.0000103000 7
SCMO1 0.000007620 0.0000007620 0.0000163000 0.0000000000 0.0000000000 0.0000068600 1
SCMO2 0.000017800 0.0000000000 0.0000235000 0.0000006340 0.0000006340 0.0000063400 2
SCMO5 0.000009560 0.0000000000 0.0000262000 0.0000005030 0.0000000000 0.0000010100 5
SCMO6 0.000006070 0.0000004140 0.0000027600 0.0000000000 0.0000000000 0.0000011000 6
SP1 0.000012700 0.0000011200 0.0000195000 0.0000071100 0.0000018700 0.0000138000 1
SP2 0.000001290 0.0000000000 0.0000018500 0.0000000000 0.0000000000 0.0000008040 2
SP3 0.000012100 0.0000089000 0.0000210000 0.0000000000 0.0000000000 0.0000000000 3
SP6 0.000005040 0.0000005040 0.0000070600 0.0000000000 0.0000000000 0.0000035300 6
SFL1 0.000017200 0.0000014300 0.0000459000 0.0000043000 0.0000017900 0.0000039400 1
SFL5 0.000009360 0.0000000000 0.0000315000 0.0000051100 0.0000017000 0.0000025500 5
SFL8 0.000013600 0.0000000000 0.0000159000 0.0000003690 0.0000007370 0.0000036900 8
VV1 0.000009760 0.0000000000 0.0000195000 0.0000004650 0.0000002320 0.0000004650 1
VV4 0.000009690 0.0000002780 0.0000026800 0.0000001850 0.0000000309 0.0000001540 4
VV5 0.000006990 0.0000000000 0.0000220000 0.0000020000 0.0000009980 0.0000049900 5
VV6 0.000004500 0.0000022500 0.0000300000 0.0000000000 0.0000000000 0.0000000000 6
VV8 0.000012200 0.0000000000 0.0000338000 0.0000000000 0.0000021500 0.0000007180 8
Vil1 0.000003290 0.0000001430 0.0000194000 0.0000014300 0.0000010700 0.0000017100 1
Vil2 0.000006480 0.0000006820 0.0000286000 0.0000003410 0.0000005970 0.0000011100 2
Vil4 0.000003720 0.0000007970 0.0000043900 0.0000001330 0.0000000000 0.0000002660 4
Vil6 0.000003460 0.0000005380 0.0000171000 0.0000003840 0.0000000768 0.0000011500 6
Vil9 0.000019300 0.0000010900 0.0000277000 0.0000037700 0.0000001450 0.0000005790 9
> pc<-PCA(don,scale.unit=TRUE,quali.sup=7)
I got the same issue yesterday as @megagrunt
@megagrunt can you tell me the result of
lapply(don, class)
> lapply(don, class)
$BAB
[1] "numeric"
$BAC
[1] "numeric"
$BAF
[1] "numeric"
$BAO
[1] "numeric"
$BAP
[1] "numeric"
$BBA
[1] "numeric"
$OP
[1] "integer"
And does it work with the following code ?
don$OP <- factor(don$OP)
pc <- PCA(don, scale.unit = TRUE, quali.sup = 7)
explor(pc)
It still does not work.
Strange... Sorry, one more demand : could you do a :
save(don, file="don.Rdata")
And send me the don.Rdata
file here ?
How do i manage to access the don.Rdata
file ?
Well, after using the save
function, it should be in your R current working directory. If you don't know where it is, you can use the getwd()
function.
Ok, this is quite strange, I just tested with your data, and I can color individual points depending on your quali.sup
variable and draw ellipses without problem...
After several tests, it doesn't work with the current CRAN version, but it does with the Github one :
devtools::install_github("juba/explor")
Are you sure it still doesn't work with the dev version for you ?
I does work actually, thank you very much !
Ah, glad to hear it. Thanks to you for helping debugging this.
@mboussaa could you try with the development version too ? I close the issue for now, but you can reopen it if it stille doesn't work.
@juba I will check that. Thank you
In fact, to use the dev version How can I proceed ?
Install the devtools
package, and then run :
devtools::install_github("juba/explor")
Perfect. Merci :)
I cannot put the confidence level on the plot using the ellispe. How is it possible?