Python 张量流：n_类问题的估计量

获取错误：

ValueError:标签形状不匹配。使用n_类=1配置的分类器。收到4份。建议的解决方法：检查n_类参数与估计器和/或标签形状。

import pandas as pd
import tensorflow as tf
import numpy as np
import os

dir_path = os.path.dirname(os.path.realpath(__file__))
csv_path = dir_path + "/good.csv"

CSV_COLUMN_NAMES = ['01', '02', '03', '04', '05', '06', '07', '08', '09', '10', '11', '12', '13', '14', '15', 'Quartile']

def load_data(y_name='Quartile'):

    all = pd.read_csv(csv_path, names=CSV_COLUMN_NAMES, header=0)

    one_hot = pd.get_dummies(all['Quartile'])
    all = all.drop('Quartile', axis=1)
    all = all.join(one_hot)

    x = all.drop([0, 1, 2, 3], axis=1)
    y = all[[0, 1, 2, 3]].copy()

    size = x.shape[0]
    cutoff = int(0.75*size)

    train_x = x.head(cutoff)
    train_y = y.head(cutoff)

    test_x = x.tail(size-cutoff)
    test_y = y.tail(size-cutoff)

    return (train_x, train_y), (test_x, test_y)

def train_input_fn(features, labels, batch_size):
    """An input function for training"""
    # Convert the inputs to a Dataset.
    dataset = tf.data.Dataset.from_tensor_slices((dict(features), labels))

    # Shuffle, repeat, and batch the examples.
    dataset = dataset.shuffle(1000).repeat().batch(batch_size)

    # Return the dataset.
    return dataset

def eval_input_fn(features, labels, batch_size):
    """An input function for evaluation or prediction"""
    features=dict(features)
    if labels is None:
        # No labels, use only features.
        inputs = features
    else:
        inputs = (features, labels)

    # Convert the inputs to a Dataset.
    dataset = tf.data.Dataset.from_tensor_slices(inputs)

    # Batch the examples
    assert batch_size is not None, "batch_size must not be None"
    dataset = dataset.batch(batch_size)

    # Return the dataset.
    return dataset

def main(argv):

    batch_size = 50;

    # Fetch the data
    (train_x, train_y), (test_x, test_y) = load_data()

    # Feature columns describe how to use the input.
    my_feature_columns = []
    for key in train_x.keys():
        my_feature_columns.append(tf.feature_column.numeric_column(key=key))

    classifier = tf.estimator.DNNClassifier(
        feature_columns=my_feature_columns,
        hidden_units=[10, 10],
        n_classes=4)

    # Train the Model.
    classifier.train(
        input_fn=lambda:train_input_fn(train_x, train_y, batch_size), steps=10)

    # Evaluate the model.
    eval_result = classifier.evaluate(
        input_fn=lambda:eval_input_fn(test_x, test_y, batch_size))

    print('\nTest set accuracy: {accuracy:0.3f}\n'.format(**eval_result))


if __name__ == '__main__':
    tf.logging.set_verbosity(tf.logging.INFO)
    tf.app.run(main)

---

我对我的输出使用了一个热编码（它是一个四分位数：通常为1-4），所以它被转换成4列，命名为：0 1 2 3。但当我去运行它时，它就好像我使用了

n_classes=1

，尽管我没有这样做。我已经做了一些关于这个问题的研究，所以不要这么快就提出重复的建议，因为这里提到的解决方案并不能解决我的问题。我没有使用mnist数据集，我使用的是自定义数据集。任何帮助都将不胜感激，谢谢

如果我没记错的话，

tf.estimator.DNNClassifier

需要一个密集的标签（比如，[2]），而不是一个热标签（比如，[0,0,1]）。因此，不要使用pd.get_dummies，并确保标签是一维数据

误导性信息已在PR中更正：