Python Keras ValueError: Data cardinality is ambiguous

I am attempting to make a model that can predict whether a credit card transaction was fraudulent or not. My dataset is available on Kaggle. Everything works up to when I fit my model, when I get this error:

ValueError: Data cardinality is ambiguous:
  x sizes: 7433462
  y sizes: 284807
Make sure all arrays contain the same number of samples.

Could someone help me figure out what’s wrong?

import numpy as np
import pandas as pd
import tensorflow as tf
from tensorflow import keras
from tensorflow.keras.models import Sequential 
from tensorflow.keras.layers import Activation, Dense
from tensorflow.keras.optimizers import Adam
from tensorflow.keras.metrics import categorical_crossentropy
from sklearn.utils import shuffle
from sklearn.preprocessing import MinMaxScaler

data = pd.read_csv("creditcard.csv")
trainSamples = data['Class']
labels = ['Time', 'V1', 'V2', 'V3', 'V4', 'V5', 'V6', 'V7', 'V8', 'V9', 'V10',  'V12', 'V13', 'V14', 'V15', 'V16',  'V17', 'V18', 'V19', 'V20', 'V21',  'V22',  'V23',  'V24',  'V25',  'V26',  'V27',  'V28',  'Amount']
trainSamples = data[labels]

trainLabels = np.array(trainLabels)
trainSamples = np.array(trainSamples)

trainLabels = shuffle(trainLabels)
trainSamples = shuffle(trainSamples)

scaler = MinMaxScaler(feature_range = (0, 1))
scaledTrainSample = scaler.fit_transform(trainSamples.reshape(-1,1))

model = Sequential([
    Dense(units = 16, input_shape = (1, ), activation = 'relu'),
    Dense(units = 32, activation = 'relu'),    
    Dense(units = 2, activation = 'softmax')
])

model.compile(optimizer = Adam(learning_rate = 0.0001), loss = 'sparse_categorical_crossentropy', metrics = ['accuracy'])
model.fit(x = scaledTrainSample, y = trainLabels, validation_split = 0.1, batch_size = 10, epochs = 300, verbose = 2)

Answer

The main issue with your code is that the model’s input shape should be 30 and not 1 as you have 30 features, while the output shape should be 1 and not 2 since you have only one binary label (i.e. only two classes, 0 or 1). There were also a few other bugs which were corrected in the code below.

import numpy as np
import pandas as pd
import tensorflow as tf
from tensorflow.keras.models import Sequential
from tensorflow.keras.layers import Dense
from tensorflow.keras.optimizers import Adam
from sklearn.utils import shuffle
from sklearn.preprocessing import MinMaxScaler
tf.random.set_seed(0)

# import the data
df = pd.read_csv('creditcard.csv')

# extract the features and target
X = df.drop(labels=['Class'], axis=1).values
y = df['Class'].values

# count the number of classes
print(np.unique(y))
# [0 1]

# shuffle the data
X, y = shuffle(X, y, random_state=42)

# scale the features
scaler = MinMaxScaler(feature_range=(0, 1))
X = scaler.fit_transform(X)

# build the model
model = Sequential([
    Dense(units=16, activation='relu', input_shape=(X.shape[1], )),
    Dense(units=32, activation='relu'),
    Dense(units=1, activation='sigmoid')
])

# fit the model
model.compile(optimizer=Adam(learning_rate=0.0001), loss='binary_crossentropy', metrics=['accuracy'])
model.fit(x=X, y=y, validation_split=0.1, batch_size=256, epochs=3)
# Epoch 1/3
# 1002/1002 [==============================] - 1s 761us/step - loss: 0.1787 - accuracy: 0.9983 - val_loss: 0.0193 - val_accuracy: 0.9981
# Epoch 2/3
# 1002/1002 [==============================] - 1s 684us/step - loss: 0.0136 - accuracy: 0.9983 - val_loss: 0.0130 - val_accuracy: 0.9981
# Epoch 3/3
# 1002/1002 [==============================] - 1s 680us/step - loss: 0.0119 - accuracy: 0.9983 - val_loss: 0.0127 - val_accuracy: 0.9981