介绍
这篇文章将为大家详细讲解有关csv文件如何实现构建Tensorflow的数据集,文章内容质量较高,因此小编分享给大家做个参考,希望大家阅读完这篇文章后对相关知识有一定的了解。
<>强基本步骤强>
- <李>获得一组csv文件的路径李> <>李将这组文件名,转成文件名对应的数据集=比;李李file_dataset > <>根据file_dataset中的每个文件名,读取文件内容生成一个内容的数据集=比;李李content_dataset > <>这样的多个content_dataset,拼接起来,形成一整个数据集李> <李>因为读出来的每条记录都是字符串类型,所以还需要对每条记录做解码
李,>
存在一个这样的变量train_filenames
pprint.pprint (train_filenames) # (& # 39;generate_csv \ \ train_00.csv& # 39;, # & # 39;generate_csv \ \ train_01.csv& # 39; # & # 39;generate_csv \ \ train_02.csv& # 39; # & # 39;generate_csv \ \ train_03.csv& # 39; # & # 39;generate_csv \ \ train_04.csv& # 39; # & # 39;generate_csv \ \ train_05.csv& # 39; # & # 39;generate_csv \ \ train_06.csv& # 39; # & # 39;generate_csv \ \ train_07.csv& # 39; # & # 39;generate_csv \ \ train_08.csv& # 39; # & # 39;generate_csv \ \ train_09.csv& # 39; # & # 39;generate_csv \ \ train_10.csv& # 39; # & # 39;generate_csv \ \ train_11.csv& # 39; # & # 39;generate_csv \ \ train_12.csv& # 39; # & # 39;generate_csv \ \ train_13.csv& # 39; # & # 39;generate_csv \ \ train_14.csv& # 39; # & # 39;generate_csv \ \ train_15.csv& # 39; # & # 39;generate_csv \ \ train_16.csv& # 39; # & # 39;generate_csv \ \ train_17.csv& # 39; # & # 39;generate_csv \ \ train_18.csv& # 39; # & # 39;generate_csv \ \ train_19.csv& # 39;]
接着,我们用提前定义好的API构建文件名数据集file_dataset
数据集=filename_dataset.interleave ( λ文件名:tf.data.TextLineDataset(文件名).skip (1), cycle_length=5 ) 在dataset.take线(3): print(线) # tf.Tensor (b # 39; 0.46908349737250216, 1.8718193706428006, 0.13936365871212536, -0.011055733363841472, -0.6349261778219746, -0.036732316700563934, 1.0259470089944995, -1.319095600336748, 2.171 & # 39;,形状=(),dtype=字符串) # tf.Tensor (b # 39; -1.102093775650278, 1.313248890578542, -0.7212003024178728, -0.14707856286537277, 0.34720121604358517, 0.0965085401826684, -0.74698820254838, 0.6810563907247876, 1.428 & # 39;,形状=(),dtype=字符串) # tf.Tensor (b # 39; -0.8901003715328659, 0.9142699762469286, -0.1851678950250224, -0.12947457252940406, 0.5958187430364827, -0.021255215877779534, 0.7914317693724252, -0.45618713536506217, 0.75 & # 39;,形状=(),dtype=string)csv文件如何实现构建Tensorflow的数据集