我在尝试使用CSV文件中的数据来训练我的神经网络，使用的是tensorflow.js，但始终没有结果。总是出现相同的错误信息（检查时出错：期望dense_Dense1_input的形状为[null,8]，但得到的数组形状为[8,1]。）。我知道有类似的问题被问过，但没有找到任何关于数据存储在CSV文件中的解答。

这是我的代码：

const dataLine = tf.tensor([0.352941,0.482412,0,0,0,0.353204,0.047822,0.116667]);columnConfigs = {outcome: {isLabel: true}};const dataset = tf.data.csv('data.csv', {columnConfigs}).map(({xs, ys}) => {return {xs:Object.values(xs), ys:Object.values(ys)}});const model = tf.sequential();model.add(tf.layers.dense({units: 12, inputShape: [8]}));model.add(tf.layers.dense({units: 1, inputShape: [12]}));model.compile({loss: 'meanSquaredError', optimizer: 'sgd'});model.fitDataset(dataset, {    epochs: 100,  });const prediction = model.predict(dataLine);prediction.print();

我还附上了一小部分我正在使用的数据：

pregnancies,glucose,blood_pressure,skin_thickness,insulin,BMI,diabetes_pedigree_function,age,outcome0.058824,0.507538,0.409836,0.151515,0.042553,0.360656,0.191289,0.083333,00.294118,0.442211,0.540984,0.212121,0.027187,0.363636,0.112724,0.15,00.470588,0.884422,0.737705,0.343434,0.35461,0.502235,0.166097,0.616667,10.411765,0.753769,0.540984,0.424242,0.404255,0.517139,0.273271,0.35,00.058824,0.366834,0.409836,0.10101,0,0.342772,0.072588,0,00.411765,0.939698,0.557377,0.393939,0.359338,0.561848,0.075149,0.333333,10,0.502513,0.721311,0.606061,0.130024,0.697466,0.377455,0.166667,00,0.733668,0.672131,0,0,0.603577,0.727156,0.383333,00,0.527638,0.52459,0.414141,0.167849,0.61848,0.040564,0.016667,00.117647,0.422111,0,0,0,0,0.096499,0,0

对此问题提供的任何帮助都将不胜感激，谢谢

回答：

尝试对数据集进行批处理：

const dataset = tf.data.csv('data.csv', {columnConfigs})    .map(({xs, ys}) => {return {xs:Object.values(xs), ys:Object.values(ys)}})    .batch(100)

然后在预测时扩展维度：

const dataLine = tf.tensor([0.352941,0.482412,0,0,0,0.353204,0.047822,0.116667])    .expandDims();...const prediction = model.predict(dataLine);