Python scikit-learn SVM 分类器“ValueError: Found array with dim 3. Expected <= 2”
声明：本页面是StackOverFlow热门问题的中英对照翻译，遵循CC BY-SA 4.0协议，如果您需要使用它，必须同样遵循CC BY-SA许可，注明原文地址和作者信息，同时你必须将它归于原作者（不是我）：StackOverFlow
原文地址: http://stackoverflow.com/questions/33162871/
Warning: these are provided under cc-by-sa 4.0 license. You are free to use/share it, But you must attribute it to the original authors (not me):
StackOverFlow
Python scikit-learn SVM Classifier "ValueError: Found array with dim 3. Expected <= 2"
提问by Hitanshu Tiwari
I am trying to implement SVM Classifier over MNIST dataset. As my parameters are 3 dimensional its throwing the following error:
我正在尝试在 MNIST 数据集上实现 SVM 分类器。由于我的参数是 3 维的，因此会引发以下错误：
ValueError: Found array with dim 3. Expected <= 2
Following is my code snippet:
以下是我的代码片段：
import mnist
from sklearn import svm
training_images, training_labels = mnist.load_mnist("training", digits = [1,2,3,4])
classifier = svm.SVC()
classifier.fit(training_images, training_labels)
Does sklearn support a multi-dimensional classifier?
sklearn 是否支持多维分类器？
采纳答案by Ryan
The problem is with your input data.
问题在于您的输入数据。
You can use sklearnto load a digit dataset as well:
您也可以使用sklearn加载数字数据集：
from sklearn.datasets import load_digits
from sklearn import svm
digits = load_digits()
X = digits.data
y = digits.target
classifier = svm.SVC()
classifier.fit(X[:1000], y[:1000])
predictions = classifier.predict(X[1000:])
回答by Zahra
One option for fixing the problem would be to reshape the input data into a 2-dimensional array.
解决该问题的一种选择是将输入数据重塑为二维数组。
Let's assume that your training data consists of 10 images which are each represented as an 3x3 matrix and therefore your input data is 3-dimensional.
假设您的训练数据由 10 张图像组成，每张图像都表示为一个 3x3 矩阵，因此您的输入数据是 3 维的。
[ [[1,2,3], [[1,2,3], [
[4,5,6], [4,5,6], image 10
[7,8,9]] , [7,8,9]] , ... , ] ]
We can turn each image into an array of 9 elements in order to convert the dataset into 2-dimensions.
我们可以将每个图像转换为 9 个元素的数组，以便将数据集转换为二维。
dataset_size = len(training_images)
TwoDim_dataset = dataset.reshape(dataset_size,-1)
This would turn the data into the following shape:
这会将数据变成以下形状：
[ [1,2,3,4,5,6,7,8,9] , [1,2,3,4,5,6,7,8,9] , ... , [image 10] ]
