Conduct principal component analysis for features extracted from keras model
Source:R/zzz_auto_visual_inference.R
AUTO_VI-cash-feature_pca.RdThis function conducts principal component analysis for features extracted from keras model.
Arguments
- feature
Dataframe. A data frame where columns represent features and rows represent observations. It should have only one row.
- null_feature
Dataframe. A data frame where columns represent features and rows represent observations. These features are extracted during the evaluation of null plots.
- boot_feature
Dataframe. A data frame where columns represent features and rows represent observations. These features are extracted during the evaluation of bootstrapped plots.
- center
Boolean. Whether to subtract the mean from the feature.
- scale
Boolean. Whether to divide the feature by its standard deviation.
- pattern
Character. A regrex pattern to search for features in the provided DataFrame. See also
grep().
Value
A tibble of the raw features and the rotated features with
attributes sdev and rotation representing the
standard deviation of the principal
components and the rotation matrix respectively.
Details
Features need to be extracted while running the method
AUTO_VI$check() and AUTO_VI$lineup_check() by providing the argument
extract_feature_from_layer. Features with zero variance will be ignored
from the analysis. See also stats::prcomp(). By default, features are
assumed to follow the naming convention
"f_(index)", where index is from one to the number of features.
Examples
keras_model <- try(get_keras_model("vss_phn_32"))
if (!inherits(keras_model, "try-error")) {
myvi <- auto_vi(lm(dist ~ speed, data = cars), keras_model)
myvi$lineup_check(extract_feature_from_layer = "global_max_pooling2d")
myvi$feature_pca()
}
#> ✔ Generate null data.
#> ✔ Generate null plots.
#> ✔ Compute auxilary inputs.
#> ✔ Predict visual signal strength for 19 images.
#> ✔ Predict visual signal strength for 1 image.
#> # A tibble: 20 × 277
#> f_1 f_2 f_3 f_4 f_5 f_6 f_7 f_8 f_9 f_10 f_11
#> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl>
#> 1 0.151 0 0 0 0 0.0203 0.109 0.0203 0 0.0834 0
#> 2 1.04 1.83 1.92 1.83 0.535 0.726 0.996 0.984 1.08 0.193 1.58
#> 3 0.204 0.414 0 0 0 0 0 0 0.421 0 0
#> 4 1.13 2.19 2.42 2.47 0.851 1.01 1.45 1.26 1.30 0.386 2.15
#> 5 0.883 1.52 1.58 1.45 0.408 0.702 0.954 0.800 0.967 0 1.28
#> 6 0.464 1.58 1.11 1.14 0.161 0.333 0.380 0.316 0.989 0.00805 0.786
#> 7 0.225 0.356 0 0.0370 0 0 0 0 0.276 0 0
#> 8 0.225 0.662 0 0.177 0 0 0 0 0.525 0 0
#> 9 0.383 0 0 0 0.232 0.242 0.359 0.344 0 0.331 0
#> 10 0.216 1.04 0.135 0.575 0 0 0 0.0894 0.507 0.0372 0.0981
#> 11 1.10 3.10 3.30 4.21 1.67 1.59 2.52 1.87 1.63 1.13 3.49
#> 12 1.03 2.40 2.42 3.01 1.26 1.15 1.80 1.52 1.19 0.940 2.41
#> 13 0.520 1.15 0.647 0.842 0 0.110 0.131 0.170 0.725 0 0.493
#> 14 0.215 0.119 0 0 0 0 0 0 0.0314 0 0
#> 15 0.429 1.07 0.615 0.704 0 0.132 0.150 0.0677 0.761 0 0.423
#> 16 0.959 2.52 2.50 2.59 0.949 1.07 1.44 1.26 1.42 0.596 2.14
#> 17 0.704 1.83 1.63 1.62 0.499 0.643 0.839 0.797 1.05 0.277 1.28
#> 18 0.806 1.91 1.78 1.73 0.514 0.673 0.871 0.861 1.11 0.264 1.41
#> 19 0.582 1.30 0.943 1.01 0.172 0.278 0.433 0.457 0.738 0.0937 0.690
#> 20 0.275 0.695 0 0.209 0 0 0 0 0.495 0 0
#> # ℹ 266 more variables: f_12 <dbl>, f_13 <dbl>, f_14 <dbl>, f_15 <dbl>,
#> # f_16 <dbl>, f_17 <dbl>, f_18 <dbl>, f_19 <dbl>, f_20 <dbl>, f_21 <dbl>,
#> # f_22 <dbl>, f_23 <dbl>, f_24 <dbl>, f_25 <dbl>, f_26 <dbl>, f_27 <dbl>,
#> # f_28 <dbl>, f_29 <dbl>, f_30 <dbl>, f_31 <dbl>, f_32 <dbl>, f_33 <dbl>,
#> # f_34 <dbl>, f_35 <dbl>, f_36 <dbl>, f_37 <dbl>, f_38 <dbl>, f_39 <dbl>,
#> # f_40 <dbl>, f_41 <dbl>, f_42 <dbl>, f_43 <dbl>, f_44 <dbl>, f_45 <dbl>,
#> # f_46 <dbl>, f_47 <dbl>, f_48 <dbl>, f_49 <dbl>, f_50 <dbl>, f_51 <dbl>, …