如何使用scikit-learn对二进制数据集进行分类？

Question

I have the following binary dataset:

[
    [1, 1, 1, 1, 1, 1, 0, 0, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 1, 1, 1, 1, 1, 1, 1, 1, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 1, 1],
    [1, 1, 1, 1, 1, 1, 0, 0, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 1, 1, 1, 1, 1, 1, 1, 1, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 1, 1],
    [1, 1, 1, 1, 1, 1, 1, 0, 1, 0, 1, 1, 1, 1, 1, 1, 1, 0, 1, 0, 1, 1, 1, 1, 1, 1, 0, 0, 1, 0, 1, 0, 1, 1, 1, 0, 0, 1, 1, 0, 1, 0, 1, 1, 1, 1, 0, 0, 1, 0],
    [1, 1, 1, 1, 1, 1, 0, 0, 1, 0, 1, 1, 1, 1, 1, 1, 0, 0, 1, 0, 1, 1, 0, 1, 0, 1, 1, 0, 1, 0, 1, 1, 1, 1, 1, 1, 0, 0, 1, 0, 1, 0, 1, 1, 1, 0, 0, 1, 1, 0],
    [1, 1, 1, 1, 1, 1, 0, 1, 1, 0, 1, 1, 1, 1, 1, 1, 0, 1, 1, 0, 1, 1, 0, 1, 1, 1, 0, 0, 1, 0, 1, 1, 0, 1, 0, 1, 1, 0, 1, 0, 1, 1, 1, 1, 1, 1, 0, 0, 1, 0]
]

I want to cluster(scatter) this into 3 parts in such a way that the arrays with most similarity(0s and 1s at the same position in array) are clustered in same group.

Since I have learned that binary data cannot be clustered straight away and need the dimensions to be reduced. The manifold learning algorithms are capable of doing that. I am trying to reduce it to as little as 2 dimensions and then scatter it on a plot to make it more user friendly, Multi-dimensional scaling seems to be the most promising algorithm for doing this. But when I fit it around my dataset, it still returns the same dataset without any reductions:

mds = MDS(n_components=2, metric=True, n_init=4, max_iter=300, verbose=0, eps=0.001, n_jobs=1, random_state=None, dissimilarity='euclidean')
mds.fit(X)
return X

Any idea what am I doing wrong or what am I missing? I am trying to reduce this data set to 2 dimensions and then cluster it on a 2d scatter graph such that the the arrays which are similar are grouped together and closely based on the similarity in positions of 0s and 1s.

Answer 1

fit() only learns the data and not changes it in any way. You need to call fit_transform() to get new transformed data. Like this:

newX = mds.fit_transform(X)

newX will be the data with 2 components as you wanted.