How to get variables of a specific component (PCA)

Question

I would like to get variables from a specific component (COMP 2) :

> pca$eig
         eigenvalue percentage of variance cumulative percentage of variance
comp 1 3.8167747447           47.709684309                          47.70968
comp 2 2.0680904354           25.851130442                          73.56081
comp 3 0.9728158313           12.160197892                          85.72101
comp 4 0.7962420036            9.953025045                          95.67404
comp 5 0.1792036107            2.240045134                          97.91408
comp 6 0.0924269941            1.155337427                          99.06942
comp 7 0.0740850743            0.926063429                          99.99548
comp 8 0.0003613058            0.004516322                         100.00000

Is there's any command in R I can use to do this?

@Waldi thank you for your comment, I use FactoMineR and factoextra — MrMR
– MrMR, Commented Mar 4, 2021 at 19:49

Waldi · Accepted Answer · 2021-03-04 20:07:31Z

2

With factoextra, you get all components with get_pca_var:

library(FactoMineR)
library(factoextra)

pca <- FactoMineR::PCA(iris[,1:4])

pca.var <- factoextra::get_pca_var(pca)

pca.var$coord

                  Dim.1      Dim.2       Dim.3       Dim.4
Sepal.Length  0.8901688 0.36082989 -0.27565767 -0.03760602
Sepal.Width  -0.4601427 0.88271627  0.09361987  0.01777631
Petal.Length  0.9915552 0.02341519  0.05444699  0.11534978
Petal.Width   0.9649790 0.06399985  0.24298265 -0.07535950

Coordinates of second dimension are:

pca.var$coord[,"Dim.2"]
Sepal.Length  Sepal.Width Petal.Length  Petal.Width 
  0.36082989   0.88271627   0.02341519   0.06399985

edited Mar 4, 2021 at 20:07

answered Mar 4, 2021 at 20:01

Waldi

41.6k6 gold badges38 silver badges90 bronze badges

Sign up to request clarification or add additional context in comments.

Comments

Manu · Accepted Answer · 2021-03-05 14:55:43Z

1

You can select an specific PC component using prcomp and the index:

library(fastDummies)
table <- data.frame(NOM = c("Evian", "Montagne des Pyrenees", "Cristaline-St-Cyr", 
                                "Fiee des Lois", "Volcania", "Saint Diery"), 
                        ACRO = c("EVIAN", "MTPYR", "CRIST", "FIEE", "VOLCA", "STDIE"), 
                        PAYS = c("F", "F","F", "F", "F", "F"), 
                        TYPE = c("M", "S", "S", "S", "S", "M"), 
                        PG = c("P", "P", "P", "P", "P", "G"), 
                        CA = c(78, 48, 71, 89, 4.1, 85), 
                        MG = c(24, 11, 5.5, 31, 1.7, 80), 
                        `NA` = c(5,34, 11.2, 17, 2.7, 385), 
                        K = c(1, 1, 3.2, 2, 0.9, 65), 
                        SUL = c(10,16, 5, 47, 1.1, 25), 
                        NO3 = c(3.8, 4, 1, 0, 0.8, 1.9), 
                        HCO3 = c(357,183, 250, 360, 25.8, 1350), 
                        CL = c(4.5, 50, 20, 28, 0.9,285), 
                        MOY = c(60.41, 43.38, 45.86, 71.75, 4.75, 284.61),
                    stringsAsFactors = TRUE)

table_conv <- dummy_cols(table,
                         select_columns = c("NOM","ACRO","PAYS","TYPE","PG"),
                         remove_selected_columns = TRUE)

pca <- prcomp(as.matrix(table_conv))

pca$x[,"PC4"]

I used your previous question as a guide for the dataset. This is your PC2 Eigenvectors:

pca$rotation[,"PC2"]
> pca$rotation[,"PC2"]
                       CA                        MG                       NA. 
             0.4583258631              0.0567007399             -0.6581371706 
                        K                       SUL                       NO3 
            -0.1004287315              0.1358452663             -0.0009273267 
                     HCO3                        CL                       MOY 
             0.3046381925             -0.4804253636             -0.0355753827 
    NOM_Cristaline-St-Cyr                 NOM_Evian         NOM_Fiee des Lois 
             0.0010557813              0.0044191573              0.0038456894 
NOM_Montagne des Pyrenees           NOM_Saint Diery              NOM_Volcania 
            -0.0029622537             -0.0016729014             -0.0046854729 
               ACRO_CRIST                ACRO_EVIAN                 ACRO_FIEE 
             0.0010557813              0.0044191573              0.0038456894 
               ACRO_MTPYR                ACRO_STDIE                ACRO_VOLCA 
            -0.0029622537             -0.0016729014             -0.0046854729 
                   PAYS_F                    TYPE_M                    TYPE_S 
             0.0000000000              0.0027462560             -0.0027462560 
                     PG_G                      PG_P 
            -0.0016729014              0.0016729014

edited Mar 5, 2021 at 14:55

answered Mar 4, 2021 at 19:56

Manu

1,13014 silver badges29 bronze badges

4 Comments

MrMR Over a year ago

Thank you, but why get_pca_var mentioned in the answer above, gives me a different result

Manu Over a year ago

Mmmm, I guess it's because the iris dataset was used. I used the dataset you posted before.

MrMR Over a year ago

Yeah, I use my data instead of the iris, and it's given me a different result, in fact, the question said : What are the variables that the second component opposes?

Manu Over a year ago

I made an edit, since you want the eigenvectors, for the second component. Please, could you edit your question and see it against the dataset I proposed? If not, we can use the classic Iris dataset.

Collectives™ on Stack Overflow

How to get variables of a specific component (PCA)

2 Answers 2

Comments

4 Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

2 Answers 2

Comments

4 Comments

Your Answer

Sign up or log in

Post as a guest

Related