I am running SHAP from the library shapper in R for a classification model intrepetation on a Keras 1D CNN model:
_____________________________________________________________________________________________________________________________________________________
Layer (type) Output Shape Param #
=====================================================================================================================================================
conv1d_8 (Conv1D) (None, 1900, 64) 384
_____________________________________________________________________________________________________________________________________________________
batch_normalization_8 (BatchNormalization) (None, 1900, 64) 256
_____________________________________________________________________________________________________________________________________________________
leaky_re_lu_8 (LeakyReLU) (None, 1900, 64) 0
_____________________________________________________________________________________________________________________________________________________
dropout_10 (Dropout) (None, 1900, 64) 0
_____________________________________________________________________________________________________________________________________________________
conv1d_9 (Conv1D) (None, 1900, 32) 22560
_____________________________________________________________________________________________________________________________________________________
batch_normalization_9 (BatchNormalization) (None, 1900, 32) 128
_____________________________________________________________________________________________________________________________________________________
leaky_re_lu_9 (LeakyReLU) (None, 1900, 32) 0
_____________________________________________________________________________________________________________________________________________________
dropout_11 (Dropout) (None, 1900, 32) 0
_____________________________________________________________________________________________________________________________________________________
conv1d_10 (Conv1D) (None, 1900, 16) 10768
_____________________________________________________________________________________________________________________________________________________
batch_normalization_10 (BatchNormalization) (None, 1900, 16) 64
_____________________________________________________________________________________________________________________________________________________
leaky_re_lu_10 (LeakyReLU) (None, 1900, 16) 0
_____________________________________________________________________________________________________________________________________________________
dropout_12 (Dropout) (None, 1900, 16) 0
_____________________________________________________________________________________________________________________________________________________
conv1d_11 (Conv1D) (None, 1900, 8) 5256
_____________________________________________________________________________________________________________________________________________________
batch_normalization_11 (BatchNormalization) (None, 1900, 8) 32
_____________________________________________________________________________________________________________________________________________________
leaky_re_lu_11 (LeakyReLU) (None, 1900, 8) 0
_____________________________________________________________________________________________________________________________________________________
dropout_13 (Dropout) (None, 1900, 8) 0
_____________________________________________________________________________________________________________________________________________________
flatten_2 (Flatten) (None, 15200) 0
_____________________________________________________________________________________________________________________________________________________
dense_4 (Dense) (None, 50) 760050
_____________________________________________________________________________________________________________________________________________________
dropout_14 (Dropout) (None, 50) 0
_____________________________________________________________________________________________________________________________________________________
dense_5 (Dense) (None, 5) 255
=====================================================================================================================================================
Total params: 799,753
Trainable params: 799,513
Non-trainable params: 240
Shap command
exp_cnn <- shap(model.CNN, data = train.conv, new_observation = test.conv)
dim(train.conv)
67 1900 1
> dim(test.conv)
26 1900 1
My train and test data are reshaped as 3D array but I am getting this error :
Error in py_call_impl(callable, dots$args, dots$keywords) : RuntimeError: Evaluation error: ValueError: Error when checking input: expected conv1d_8_input to have 3 dimensions, but got array with shape (67, 1).
I am getting the same error when using individual_variable_effect function instead os shap function.