1

I'm just wondering how to access the percentage of variation explained and cross-validated test error after using a PCR regression. I've looked at the documentation for summary.mvr, but I think I must be missing something. Is there an easy way to do it? For example:

library(MASS)
boston = Boston

pcr.fit = pcr(crim ~ ., data = boston, scale = TRUE, validation = "CV")
summary(pcr.fit)

Data:   X dimension: 506 13 
    Y dimension: 506 1
Fit method: svdpc
Number of components considered: 13

VALIDATION: RMSEP
Cross-validated using 10 random segments.
       (Intercept)  1 comps  2 comps  3 comps  4 comps  5 comps  6 comps  7 comps  8 comps  9 comps
CV            8.61    7.174    7.178    6.722    6.703    6.717    6.727    6.729    6.603    6.611
adjCV         8.61    7.173    7.177    6.719    6.699    6.715    6.723    6.724    6.598    6.607
       10 comps  11 comps  12 comps  13 comps
CV        6.624     6.643     6.569     6.498
adjCV     6.619     6.635     6.561     6.490

TRAINING: % variance explained
      1 comps  2 comps  3 comps  4 comps  5 comps  6 comps  7 comps  8 comps  9 comps  10 comps
X       47.70    60.36    69.67    76.45    82.99    88.00    91.14    93.45    95.40     97.04
crim    30.69    30.87    39.27    39.61    39.61    39.86    40.14    42.47    42.55     42.78
      11 comps  12 comps  13 comps
X        98.46     99.52     100.0
crim     43.04     44.13      45.4

I would simply just like to access the percentage of variation explained for the predictors and crim. Is there an easy way to do this?

Thank you for all your help!

Vincent

Vincent
  • 7,808
  • 13
  • 49
  • 63

2 Answers2

1

To get the percentage of variation explained for each predictors, you can use:

cumsun(explvar(pcr.fit))

For the cross-validated test errors, you can maybe look at the MSEP function.

Also if in this case you can use getS3method("summary","mvr") to see the code of the summary function and figure out how to get the values you need from there.

NicE
  • 21,165
  • 3
  • 51
  • 68
0

For anyone else who stumbles upon this question, here's the command (post pcr command as in the initial question) that worked for me to plot the number of components against the total explained variation. 

validationplot(pcr.fit, val.type = "R2")
APH
  • 4,109
  • 1
  • 25
  • 36