实际情况是：在读取文件中的“类别”信息，将类别信息转为id，且顺序按照原来的数据顺序排序

labels = list(data['product_classification'])
unordered = set(labels)
print("unorderd_class: ",unordered)
order = sorted(unordered, key=labels.index)
print("ordered_class: ", order)
label_ = dict(zip(order, range(len(order))))
labels_true = data["product_classification"].map(label_)