AttributeError: 'numpy.ndarray' object has no attribute 'id'

问题

I am creating a sklearn pipeline that consists of 3 steps:

Transforms pandas dataframe into 3D array
Transforms 3D array into recurrence plot (image)
Trains an image classification model using Keras

This is my initial data set:

train_df - pandas dataframe

id   cycle   s1
1    1       0.05
1    2       0.04
1    3       0.05
1    4       0.05
2    1       0.02
2    2       0.03

y_train

array([[1., 0., 0.],
       [1., 0., 0.],
       ...
       [1., 0., 0.]], dtype=float32)

When I run my current code (see below), I get the following error:

AttributeError: 'numpy.ndarray' object has no attribute 'id'

Code (As you may notice I only transform X in the pipeline. Ideally I would like to transform y in the pipeline as well):

import numpy as np
from keras.wrappers.scikit_learn import KerasClassifier
import tensorflow as tf
from tensorflow.keras.models import *      # I know that this is not a best way to import libraries
from tensorflow.keras.layers import *      # due to possible issues with namespace
from tensorflow.keras.optimizers import *  # But my current problem is not related to this aspect
from tensorflow.keras.utils import *       # Therefore please ignore it.
from tensorflow.keras.callbacks import *
from sklearn.pipeline import Pipeline

def create_model():
    model = Sequential()

    model.add(Conv2D(32, (3, 3), activation='relu', input_shape=(50, 50, 17)))
    model.add(Conv2D(32, (3, 3), activation='relu'))
    model.add(MaxPooling2D(pool_size=(2, 2)))
    model.add(Dropout(0.25))

    model.add(Conv2D(64, (3, 3), activation='relu'))
    model.add(Conv2D(64, (3, 3), activation='relu'))
    model.add(MaxPooling2D(pool_size=(2, 2)))
    model.add(Dropout(0.25))

    model.add(Flatten())
    model.add(Dense(256, activation='relu'))
    model.add(Dropout(0.5))
    model.add(Dense(3, activation='softmax'))

    model.compile(loss='categorical_crossentropy', optimizer='adam', metrics=['accuracy'])

    print(model.summary())
    
    return model


def _rec_plot(s, eps=0.10, steps=10):
    d = pdist(s[:,None])
    d = np.floor(d/eps)
    d[d>steps] = steps
    Z = squareform(d)
    return Z


def 2Dto3D(X, y=None):
    x = []
    for engine_id in X.id.unique():
        for sequence in gen_sequence(X[X.id==engine_id], sequence_length, sequence_cols):
            x.append(sequence)

    x = np.asarray(x)
    return x


def ArrayToImage(x, y=None):
    x = np.apply_along_axis(_rec_plot, 1, x).astype('float16')  
    return x

transformer1 = sklearn.preprocessing.FunctionTransformer(func=2Dto3D)
transformer2 = sklearn.preprocessing.FunctionTransformer(func=ArrayToImage)

clf = KerasClassifier(build_fn=create_model, verbose=0)

blackbox_model = Pipeline([('2Dto3D', transformer1),('3DtoImage', transformer2),('clf',clf)])
blackbox_model.fit(train_df, y_train)

来源：https://stackoverflow.com/questions/63330277/attributeerror-numpy-ndarray-object-has-no-attribute-id

标签

python

keras

scikit-learn

pipeline

tf.keras