Correspondence Analysis in Rstudio

Hey guys,
I have a data set that is made of qualitatives variables that I'm trying to correlate. To that end I'm trying to perform a multiple correspendence analysis with the package CA, and the function mjca
Here is what my data look like : N= NO, O= Yes

> head(data3,20)
   dens_ville         climat      prov promo sub_mail GDP_ca benef_quali
1    moderate cold semi-arid Instagram     N        O     60k           +
2    very_low cold semi-arid    direct     N        O     40k           +
3    moderate cold semi-arid Instagram     O        O     60k           -
4         low cold semi-arid  Facebook     N        O     60k           +
5    very_low cold semi-arid  Facebook     N        O     60k           +
6    moderate  hot semi-arid  Facebook     N        O     60k           +
7    moderate  hot semi-arid Instagram     N        O     60k           -
8    moderate    continental  Facebook     N        O     50k           +
9    moderate    continental    direct     N        N     50k           +
10   moderate    continental Instagram     N        N     50k           +
11   moderate    continental    direct     N        N     50k           +
12   moderate    continental Instagram     N        O     50k           -
13   moderate    continental  Facebook     N        N     50k           -
14        low    continental    direct     O        O     60k           -
15        low    continental    direct     N        O     50k           +
16   moderate    continental Instagram     O        O     50k           +
17   moderate    continental Instagram     O        O     50k           +
18   moderate    continental  Facebook     N        N     60k           -
19   moderate    continental Instagram     N        N     60k           -
20        low    continental Instagram     N        N     50k           -

My Goal is : I'm looking to see what correlates with benefice +.
My Question is : which map type should I chose for plotting (and why)?
I not too familiar with this method and I've read the paper by Nenadié & Greenacre, but that doesn't really tell me which map type I should chose for the plot. (I'm relatively new to statistics btw)

Rowgab or Rowgreen seem the only readables ones, but that doesn't mean they are actually appropriate vs the default symmetrical one.
Cheers

PS:
If you know of a better analysis to do what I want to do, please feel free to suggest it !
I have tried randomforest regression and logistic regression with poor success (for those, all numbered values (density, GDP, benef) were not converted to qualitative (except benef for logistic regression).

Preparatory:

  1. replace - with _ in variable names
  2. replace - with minus and + with plus for last variable
  3. convert all variables to factors
  4. Review mjca function signature for expected argument

ob j A response pattern matrix (data frame containing factors), or a frequency table(a “table” object) or an integer array

suppressPackageStartupMessages({
  library(ca)
})

input <- as.data.frame(structure(list(
  dens_ville = c(
    "moderate", "very_low", "moderate",
    "low", "very_low", "moderate", "moderate", "moderate", "moderate",
    "moderate", "moderate", "moderate", "moderate", "low", "low",
    "moderate", "moderate", "moderate", "moderate", "low"
  ), climat = c(
    "cold_semi_arid",
    "cold_semi_arid", "cold_semi_arid", "cold_semi_arid", "cold_semi_arid",
    "hot_semi_arid", "hot_semi_arid", "continental", "continental",
    "continental", "continental", "continental", "continental", "continental",
    "continental", "continental", "continental", "continental", "continental",
    "continental"
  ), prov = c(
    "Instagram", "direct", "Instagram",
    "Facebook", "Facebook", "Facebook", "Instagram", "Facebook",
    "direct", "Instagram", "direct", "Instagram", "Facebook", "direct",
    "direct", "Instagram", "Instagram", "Facebook", "Instagram",
    "Instagram"
  ), promo = c(
    "N", "N", "O", "N", "N", "N", "N", "N",
    "N", "N", "N", "N", "N", "O", "N", "O", "O", "N", "N", "N"
  ),
  sub_mail = c(
    "O", "O", "O", "O", "O", "O", "O", "O", "N",
    "N", "N", "O", "N", "O", "O", "O", "O", "N", "N", "N"
  ), GDP_ca = c(
    "60k",
    "40k", "60k", "60k", "60k", "60k", "60k", "50k", "50k", "50k",
    "50k", "50k", "50k", "60k", "50k", "50k", "50k", "60k", "60k",
    "50K"
  ), benef_quali = c(
    "plus", "plus", "minus", "plus",
    "plus", "plus", "minus", "plus", "plus", "plus", "plus",
    "minus", "minus", "minus", "plus", "plus", "plus", "minus",
    "minus", "minus"
  )
), class = c(
  "spec_tbl_df", "tbl_df", "tbl",
  "data.frame"
), row.names = c(NA, -20L), spec = structure(list(
  cols = list(dens_ville = structure(list(), class = c(
    "collector_character",
    "collector"
  )), climat = structure(list(), class = c(
    "collector_character",
    "collector"
  )), prov = structure(list(), class = c(
    "collector_character",
    "collector"
  )), promo = structure(list(), class = c(
    "collector_character",
    "collector"
  )), sub_mail = structure(list(), class = c(
    "collector_character",
    "collector"
  )), GDP_ca = structure(list(), class = c(
    "collector_character",
    "collector"
  )), benef_quali = structure(list(), class = c(
    "collector_character",
    "collector"
  ))), default = structure(list(), class = c(
    "collector_guess",
    "collector"
  )), skip = 1L
), class = "col_spec")))

mjca(input)
#> 
#>  Eigenvalues:
#>            1        2        3       4        5      
#> Value      0.079674 0.025836 0.01188 0.008318 0.00043
#> Percentage 45.01%   14.59%   6.71%   4.7%     0.24%  
#> 
#> 
#>  Columns:
#>         dens_ville:low dens_ville:moderate dens_ville:very_low
#> Mass          0.028571            0.100000            0.014286
#> ChiDist       0.887074            0.326054            1.640732
#> Inertia       0.022483            0.010631            0.038457
#> Dim. 1        0.203258            0.508740           -3.967700
#> Dim. 2        0.521807           -0.313550            1.151234
#>         climat:cold_semi_arid climat:continental climat:hot_semi_arid
#> Mass                 0.035714           0.092857             0.014286
#> ChiDist              0.926191           0.406944             1.302186
#> Inertia              0.030637           0.015377             0.024224
#> Dim. 1              -2.184247           0.884040            -0.285640
#> Dim. 2              -0.612577           0.750688            -3.348028
#>         prov:direct prov:Facebook prov:Instagram   promo:N   promo:O sub_mail:N
#> Mass       0.035714      0.042857       0.064286  0.114286  0.028571   0.050000
#> ChiDist    0.796396      0.654224       0.500775  0.220038  0.880154   0.680510
#> Inertia    0.022652      0.018343       0.016121  0.005533  0.022133   0.023155
#> Dim. 1    -0.611919     -0.527934       0.691910 -0.078394  0.313577   1.246810
#> Dim. 2     2.270584     -0.831213      -0.707294  0.131480 -0.525919   0.899438
#>         sub_mail:O GDP_ca:40k GDP_ca:50k GDP_ca:50K GDP_ca:60k
#> Mass      0.092857   0.007143   0.064286   0.007143   0.064286
#> ChiDist   0.366428   2.250560   0.557181   2.013094   0.535171
#> Inertia   0.012468   0.036179   0.019958   0.028947   0.018412
#> Dim. 1   -0.671359  -4.783702   0.752008   2.261258  -0.471736
#> Dim. 2   -0.484313   3.164784   1.211729   0.728415  -1.644307
#>         benef_quali:minus benef_quali:plus
#> Mass             0.057143         0.085714
#> ChiDist          0.569845         0.379897
#> Inertia          0.018556         0.012370
#> Dim. 1           0.882021        -0.588014
#> Dim. 2          -0.893626         0.595750

Created on 2020-12-04 by the reprex package (v0.3.0.9001)

1 Like

This topic was automatically closed 21 days after the last reply. New replies are no longer allowed.

If you have a query related to it or one of the replies, start a new topic and refer back with a link.

Hi, thanks for your reply, I haven't had time to explore your suggestion yet as I am in the middle of my midterms.
I will try to implement it and get back to you when I have some time on my hands.
Cheers