ValueError：model 层需要 21 个输入，但它接收到 1 个输入张量

Question

I am beginner to Keras. I am following this example where I am training binary classification model using Keras where input data is structured data taken from a csv, but I am getting following error

ValueError: Layer model expects 21 input(s), but it received 1 input tensors. Inputs received: [<tf.Tensor 'ExpandDims:0' shape=(None, 1) dtype=float64>]

at line

score = model.evaluate(x=test_labels, y=test_data, verbose=1)

My code look like below

import os
import numpy as np
import pandas as pd
import tensorflow as tf
from tensorflow import keras
from tensorflow.keras import layers
from tensorflow.keras.layers.experimental.preprocessing import Normalization

def dataframe_to_dataset(dataframe):
    dataframe = dataframe.copy()
    labels = dataframe.pop("label")
    labels = np.asarray(labels).astype('float32').reshape((-1, 1))
    ds = tf.data.Dataset.from_tensor_slices((dict(dataframe), labels))
    ds = ds.shuffle(buffer_size=len(dataframe))
    return ds


os.chdir('datasets/accelerometer/accel-labeled/')
df = pd.read_csv("combined-all-labeled.csv", delimiter=',')

print(f"All size : {df.shape[0]}")

np.random.seed(23)
perm = np.random.permutation(df.index)
m = len(df.index)
train_end = int(.70 * m)
validate_end = int(.25 * m) + train_end

train_ds = dataframe_to_dataset(df.iloc[perm[:train_end]])
validate_ds = dataframe_to_dataset(df.iloc[perm[train_end:validate_end]])
test_df = df.iloc[perm[validate_end:]]

test_labels = test_df['label'].astype('float')
test_data = test_df.iloc[:, 2:22].astype('float')
print(test_data.head(2))

print(f"Train Set size : {len(train_ds)}")
print(f"Validation Set size : {len(validate_ds)}")
print(f"Test Set size : {len(test_df)}")


def encode_numerical_feature(feature, name, dataset):
    # Create a Normalization layer for our feature
    normalizer = Normalization()

    # Prepare a Dataset that only yields our feature
    feature_ds = dataset.map(lambda x, y: x[name])
    feature_ds = feature_ds.map(lambda x: tf.expand_dims(x, -1))

    # Learn the statistics of the data
    normalizer.adapt(feature_ds)

    # Normalize the input feature
    encoded_feature = normalizer(feature)
    return encoded_feature


# Numerical features
x_mean = keras.Input(shape=(1,), name="x_mean")
x_median = keras.Input(shape=(1,), name="x_median")
x_std_dev = keras.Input(shape=(1,), name="x_stdev")
x_raw_min = keras.Input(shape=(1,), name="x_raw_min")
x_raw_max = keras.Input(shape=(1,), name="x_raw_max")
x_abs_min = keras.Input(shape=(1,), name="x_abs_min")
x_abs_max = keras.Input(shape=(1,), name="x_abs_max")

y_mean = keras.Input(shape=(1,), name="y_mean")
y_median = keras.Input(shape=(1,), name="y_median")
y_std_dev = keras.Input(shape=(1,), name="y_stdev")
y_raw_min = keras.Input(shape=(1,), name="y_raw_min")
y_raw_max = keras.Input(shape=(1,), name="y_raw_max")
y_abs_min = keras.Input(shape=(1,), name="y_abs_min")
y_abs_max = keras.Input(shape=(1,), name="y_abs_max")

z_mean = keras.Input(shape=(1,), name="z_mean")
z_median = keras.Input(shape=(1,), name="z_median")
z_std_dev = keras.Input(shape=(1,), name="z_stdev")
z_raw_min = keras.Input(shape=(1,), name="z_raw_min")
z_raw_max = keras.Input(shape=(1,), name="z_raw_max")
z_abs_min = keras.Input(shape=(1,), name="z_abs_min")
z_abs_max = keras.Input(shape=(1,), name="z_abs_max")

all_inputs = [
    x_mean,
    x_median,
    x_std_dev,
    x_raw_min,
    x_raw_max,
    x_abs_min,
    x_abs_max,
    y_mean,
    y_median,
    y_std_dev,
    y_raw_min,
    y_raw_max,
    y_abs_min,
    y_abs_max,
    z_mean,
    z_median,
    z_std_dev,
    z_raw_min,
    z_raw_max,
    z_abs_min,
    z_abs_max,
]

# Numerical features
x_mean_encoded = encode_numerical_feature(x_mean, "x_mean", train_ds)
x_median_encoded = encode_numerical_feature(x_median, "x_median", train_ds)
x_std_dev_encoded = encode_numerical_feature(x_std_dev, "z_stdev", train_ds)
x_raw_min_encoded = encode_numerical_feature(x_raw_min, "x_raw_min", train_ds)
x_raw_max_encoded = encode_numerical_feature(x_raw_max, "x_raw_max", train_ds)
x_abs_min_encoded = encode_numerical_feature(x_abs_min, "x_abs_min", train_ds)
x_abs_max_encoded = encode_numerical_feature(x_abs_max, "x_abs_max", train_ds)

y_mean_encoded = encode_numerical_feature(y_mean, "y_mean", train_ds)
y_median_encoded = encode_numerical_feature(y_median, "y_median", train_ds)
y_std_dev_encoded = encode_numerical_feature(y_std_dev, "z_stdev", train_ds)
y_raw_min_encoded = encode_numerical_feature(y_raw_min, "y_raw_min", train_ds)
y_raw_max_encoded = encode_numerical_feature(y_raw_max, "y_raw_max", train_ds)
y_abs_min_encoded = encode_numerical_feature(y_abs_min, "y_abs_min", train_ds)
y_abs_max_encoded = encode_numerical_feature(y_abs_max, "y_abs_max", train_ds)

z_mean_encoded = encode_numerical_feature(z_mean, "z_mean", train_ds)
z_median_encoded = encode_numerical_feature(z_median, "z_median", train_ds)
z_std_dev_encoded = encode_numerical_feature(z_std_dev, "z_stdev", train_ds)
z_raw_min_encoded = encode_numerical_feature(z_raw_min, "z_raw_min", train_ds)
z_raw_max_encoded = encode_numerical_feature(z_raw_max, "z_raw_max", train_ds)
z_abs_min_encoded = encode_numerical_feature(z_abs_min, "z_abs_min", train_ds)
z_abs_max_encoded = encode_numerical_feature(z_abs_max, "z_abs_max", train_ds)

all_features = layers.concatenate(
    [
        x_mean_encoded,
        x_median_encoded,
        x_std_dev_encoded,
        x_raw_min_encoded,
        x_raw_max_encoded,
        x_abs_min_encoded,
        x_abs_max_encoded,

        y_mean_encoded,
        y_median_encoded,
        y_std_dev_encoded,
        y_raw_min_encoded,
        y_raw_max_encoded,
        y_abs_min_encoded,
        y_abs_max_encoded,

        z_mean_encoded,
        z_median_encoded,
        z_std_dev_encoded,
        z_raw_min_encoded,
        z_raw_max_encoded,
        z_abs_min_encoded,
        z_abs_max_encoded,
    ]
)

x = layers.Dense(32, activation="relu")(all_features)
x = layers.Dropout(0.5)(x)
output = layers.Dense(1, activation="sigmoid")(x)
model = keras.Model(all_inputs, output)
model.compile("adam", "binary_crossentropy", metrics=["accuracy"])

# `rankdir='LR'` is to make the graph horizontal.
keras.utils.plot_model(model, show_shapes=True, rankdir="LR")

# model.summary()

# Train model
model.fit(train_ds, epochs=1, validation_data=validate_ds)

score = model.evaluate(x=test_labels, y=test_data, verbose=1)
#
print('Test loss: ', score[0])
print('Test accuracy: ', score[1])

CSV look like below

group_timestamp,label,x_mean,x_median,x_stdev,x_raw_min,x_raw_max,x_abs_min,x_abs_max,y_mean,y_median,y_stdev,y_raw_min,y_raw_max,y_abs_min,y_abs_max,z_mean,z_median,z_stdev,z_raw_min,z_raw_max,z_abs_min,z_abs_max
2017-05-02 17:35:20,0,-8.40793368,-8.432378499999999,0.0812278134949539,-8.632295,-8.24563,8.24563,8.632295,-180900768.0,-180900768.0,0.0,-180900768.0,-180900768.0,180900768.0,180900768.0,180900768.0,180900768.0,0.0,180900768.0,180900768.0,180900768.0,180900768.0
2017-05-02 17:19:40,0,1.96025263,1.9716251,0.0710845152401064,1.816002,2.112883,1.816002,2.112883,-180900768.0,-180900768.0,0.0,-180900768.0,-180900768.0,180900768.0,180900768.0,180900752.0,180900752.0,0.0,180900752.0,180900752.0,180900752.0,180900752.0
...

How could I fix this?

---

Additionally any other input fix/improve code/model are welcome!

Answer 1

Improvement suggestion, check the command in line 174: