Skip to contents

Linear Discriminant Analysis on Coe objects

Source: R/mult-LDA.R
LDA.Rd

Calculates a LDA on Coe on top of MASS::lda.

Usage

LDA(x, fac, retain, ...)

# S3 method for default
LDA(x, fac, retain, ...)

# S3 method for PCA
LDA(x, fac, retain = 0.99, ...)

Arguments

x

a Coe or a PCA object

fac

the grouping factor (names of one of the $fac column or column id)

retain

the proportion of the total variance to retain (if retain<1) using scree, or the number of PC axis (if retain>1).

...

additional arguments to feed lda

Value

a 'LDA' object on which to apply plot.LDA, which is a list with components:

  • x any Coe object (or a matrix)

  • fac grouping factor used

  • removed ids of columns in the original matrix that have been removed since constant (if any)

  • mod the raw lda mod from lda

  • mod.pred the predicted model using x and mod

  • CV.fac cross-validated classification

  • CV.tab cross-validation tabke

  • CV.correct proportion of correctly classified individuals

  • CV.ce class error

  • LDs unstandardized LD scores see Claude (2008)

  • mshape mean values of coefficients in the original matrix

  • method inherited from the Coe object (if any)

Note

For LDA.PCA, retain can be passed as a vector (eg: 1:5, and retain=1, retain=2, ..., retain=5) will be tried, or as "best" (same as before but retain=1:number_of_pc_axes is used).

Silent message and progress bars (if any) with options("verbose"=FALSE).

See also

Other multivariate: CLUST(), KMEANS(), KMEDOIDS(), MANOVA_PW(), MANOVA(), MDS(), MSHAPES(), NMDS(), PCA(), classification_metrics()

Examples

bot.f <- efourier(bot, 24)
#> 'norm=TRUE' is used and this may be troublesome. See ?efourier #Details
bot.p <- PCA(bot.f)
LDA(bot.p, 'type', retain=0.99) # retains 0.99 of the total variance
#> 8 PC retained
#>  * Cross-validation table ($CV.tab):
#>         classified
#> actual   beer whisky
#>   beer     17      3
#>   whisky    1     19
#> 
#>  * Class accuracy ($CV.ce):
#>   beer whisky 
#>   0.85   0.95 
#> 
#>  * Leave-one-out cross-validation ($CV.correct): (90% - 36/40): 
LDA(bot.p, 'type', retain=5) # retain 5 axis
#> 5 PC retained
#>  * Cross-validation table ($CV.tab):
#>         classified
#> actual   beer whisky
#>   beer     16      4
#>   whisky    4     16
#> 
#>  * Class accuracy ($CV.ce):
#>   beer whisky 
#>    0.8    0.8 
#> 
#>  * Leave-one-out cross-validation ($CV.correct): (80% - 32/40): 
bot.l <- LDA(bot.p, 'type', retain=0.99)
#> 8 PC retained
plot_LDA(bot.l)
#> * Only two levels, so a single LD and preparing for an histogram

#> $xy
#>                        LD1
#> brahma          2.06882655
#> caney           1.95733171
#> chimay          3.18567319
#> corona          1.91972111
#> deusventrue     1.51983847
#> duvel           3.25459981
#> franziskaner    1.20540643
#> grimbergen      1.78612198
#> guiness         0.31717542
#> hoegardeen      2.23601856
#> jupiler         2.41738081
#> kingfisher      1.17563178
#> latrappe        2.48017277
#> lindemanskriek  0.84132717
#> nicechouffe    -0.20973451
#> pecheresse      2.73987210
#> sierranevada    2.12878315
#> tanglefoot      0.50802841
#> tauro           2.45585085
#> westmalle       1.86373375
#> amrut          -1.63004033
#> ballantines    -3.31173062
#> bushmills      -1.09107572
#> chivas         -1.97923449
#> dalmore        -0.60822705
#> famousgrouse   -1.43517709
#> glendronach    -1.56712869
#> glenmorangie   -1.46854222
#> highlandpark   -2.62231929
#> jackdaniels    -0.70483285
#> jb             -2.24638367
#> johnniewalker  -0.97832954
#> magallan       -2.17603623
#> makersmark     -1.23316404
#> oban           -2.34336958
#> oldpotrero     -0.37639098
#> redbreast      -4.17639973
#> tamdhu         -0.09818091
#> wildturkey     -3.48193578
#> yoichi         -2.32326072
#> 
#> $f
#>  type1  type2  type3  type4  type5  type6  type7  type8  type9 type10 type11 
#> whisky whisky whisky whisky whisky whisky whisky whisky whisky whisky whisky 
#> type12 type13 type14 type15 type16 type17 type18 type19 type20 type21 type22 
#> whisky whisky whisky whisky whisky whisky whisky whisky whisky   beer   beer 
#> type23 type24 type25 type26 type27 type28 type29 type30 type31 type32 type33 
#>   beer   beer   beer   beer   beer   beer   beer   beer   beer   beer   beer 
#> type34 type35 type36 type37 type38 type39 type40 
#>   beer   beer   beer   beer   beer   beer   beer 
#> Levels: beer whisky
#> 
#> $colors_groups
#> [1] "#66C2A5FF" "#FC8D62FF"
#> 
#> $colors_rows
#>  [1] "#FC8D62FF" "#FC8D62FF" "#FC8D62FF" "#FC8D62FF" "#FC8D62FF" "#FC8D62FF"
#>  [7] "#FC8D62FF" "#FC8D62FF" "#FC8D62FF" "#FC8D62FF" "#FC8D62FF" "#FC8D62FF"
#> [13] "#FC8D62FF" "#FC8D62FF" "#FC8D62FF" "#FC8D62FF" "#FC8D62FF" "#FC8D62FF"
#> [19] "#FC8D62FF" "#FC8D62FF" "#66C2A5FF" "#66C2A5FF" "#66C2A5FF" "#66C2A5FF"
#> [25] "#66C2A5FF" "#66C2A5FF" "#66C2A5FF" "#66C2A5FF" "#66C2A5FF" "#66C2A5FF"
#> [31] "#66C2A5FF" "#66C2A5FF" "#66C2A5FF" "#66C2A5FF" "#66C2A5FF" "#66C2A5FF"
#> [37] "#66C2A5FF" "#66C2A5FF" "#66C2A5FF" "#66C2A5FF"
#> 
#> $object
#> [1] "PCA"
#> 
#> $axes
#> [1] 1
#> 
#> $palette
#> function (n, transp = 0) 
#> {
#>     .pal_brewer(n, "Set2") %>% pal_alpha(transp = transp)
#> }
#> <bytecode: 0x55f56426dfb0>
#> <environment: namespace:Momocs>
#> 
#> $method
#> [1] "LDAPCA"
#> 
#> $mshape
#> NULL
#> 
#> $cuts
#> NULL
#> 
#> $eig
#> NULL
#> 
#> $sdev
#> [1] 11.33732
#> 
#> $rotation
#>               LD1
#> PC1 -0.0021966326
#> PC2 -0.0004659024
#> PC3  0.0028778651
#> PC4 -0.0020601963
#> PC5  0.0016842402
#> PC6 -0.0008112386
#> PC7 -0.0006776597
#> PC8 -0.0004229307
#> 
#> $LDs
#>               LD1
#> PC1 -0.0021966326
#> PC2 -0.0004659024
#> PC3  0.0028778651
#> PC4 -0.0020601963
#> PC5  0.0016842402
#> PC6 -0.0008112386
#> PC7 -0.0006776597
#> PC8 -0.0004229307
#> 
#> $baseline1
#> NULL
#> 
#> $baseline2
#> NULL
#> 
#> $links
#> NULL
#> 
bot.f <- mutate(bot.f, plop=factor(rep(letters[1:4], each=10)))
bot.l <- LDA(PCA(bot.f), 'plop')
#> 8 PC retained
plot_LDA(bot.l) # will replace the former soon