TensorFlow 2.0 中神经网络的问题

Question

import tensorflow as tf

import pandas as pd

import numpy as np

import matplotlib as plt

from sklearn.model_selection import train_test_split

from tensorflow import keras

from tensorflow.keras.models import Sequential

from tensorflow.keras.layers import Dense

from sklearn.preprocessing import StandardScaler

import functools

LABEL_COLUMN = 'Endstage'

LABELS = [1, 2, 3, 4]

x = pd.read_csv('HCVnew.csv', index_col=False)

def get_dataset(file_path, **kwargs):

  dataset = tf.data.experimental.make_csv_dataset(

      file_path,

      batch_size=35, # Artificially small to make examples easier to show.

      label_name=LABEL_COLUMN,

      na_value="?",

      num_epochs=1,

      ignore_errors=True,

      **kwargs)

  return dataset

SELECT_COLUMNS = ["Alter", "Gender", "BMI", "Fever", "Nausea", "Fatigue",

                  "WBC", "RBC", "HGB", "Plat", "AST1", "ALT1", "ALT4", "ALT12", "ALT24", "ALT36", "ALT48", "ALT24w",

                  "RNABase", "RNA4", "Baseline", "Endstage"]

DEFAULTS = [0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0,

            0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0]

temp_dataset = get_dataset("HCVnew.csv",

                           select_columns=SELECT_COLUMNS,

                           column_defaults=DEFAULTS)

def pack(features, label):

  return tf.stack(list(features.values()), axis=-1), label

packed_dataset = temp_dataset.map(pack)

"""

for features, labels in packed_dataset.take(1):

  print(features.numpy())

  print()

  print(labels.numpy())

"""

NUMERIC_FEATURES = ["Alter", "Gender","BMI", "Fever", "Nausea", "Fatigue",

                  "WBC", "RBC", "HGB", "Plat", "AST1", "ALT1", "ALT4", "ALT12", "ALT24", "ALT36", "ALT48", "ALT24w",

                  "RNABase", "RNA4", "Baseline", "Endstage"]

您好，我正在尝试构建一个神经网络，该神经网络可以根据包含患者信息的csv文件预测丙型肝炎，我无法修复错误...我收到错误：KeyError'Endstage'，而Endstage是包含相应值（介于1和4之间）并用作标签列的csv列。如果有人有一个想法可以解决我的问题，那么请告诉我。非常感谢您的帮助！

Answer 1

这是因为是你的标签列，框架通过从数据集中删除（弹出）它来帮你一个忙。否则，您的训练数据集也将具有目标类，使其无用。Endstage将其从任何其他位置删除，使其进入训练集功能。NUMERIC_FEATURES[编辑]OP在后续问题（在评论中）中询问了为什么在修复初始问题后，他会遇到错误：值错误：特征数字不在特征字典中从外观上看，调用的特征是通过调用 来生成的。后者用于创建 和 ，但这些从不使用。然而，这句话：numericPackNumericFeaturespacked_train_datapacked_test_datanumeric_column = tf.feature_column.numeric_column('numeric', normalizer_fn=normalizer, shape=[len(NUMERIC_FEATURES)])假设数据在那里 - 因此出现错误。