【t-SNE可视化-代码】

代码

整体代码

t-SNE.py

import numpy as np
import matplotlib.pyplot as plt
from sklearn import datasets
from sklearn.manifold import TSNE

PATH = 'D:/Code/test1/Vector/'
# PATH = 'D:/Code/Vector_mean/'

vector = []

my_label = []

# 加载数据
def get_data():
    """
    :return: 数据集、标签、样本数量、特征数量
    """
    digits = datasets.load_digits(n_class=10)
    data = digits.data		# 图片特征
    label = digits.target		# 图片标签
    n_samples, n_features = data.shape		# 数据集的形状
    return data, label, n_samples, n_features

def get_mydata():
    class_name = ['CatVector.txt', 'DogVector.txt']
    class_label = ['Cat', 'Dog']
    for i in range(len(class_name)):
        vector = np.loadtxt(PATH + class_name[i])
        # vector = np.expand_dims(vector, axis = 0)
        for j in range(vector.shape[0]):
            my_label.append(class_label[i])
        if i == 0:
            allvector = vector
        else:
            allvector = np.concatenate((allvector, vector), axis=0)

    print(allvector.shape)
    print(np.array(my_label).shape)

    return allvector, np.array(my_label)
# 对样本进行预处理并画图
def plot_embedding(data, label, title):
    """
    :param data:数据集
    :param label:样本标签
    :param title:图像标题
    :return:图像
    """
    x_min, x_max = np.min(data, 0), np.max(data, 0)
    data = (data - x_min) / (x_max - x_min)		# 对数据进行归一化处理
    fig = plt.figure()		# 创建图形实例
    ax = plt.subplot(111)		# 创建子图
    # 遍历所有样本
    for i in range(data.shape[0]):
        # 在图中为每个数据点画出标签
        if label[i] == 'Cat':
            # colorplt = plt.cm.Set1(0.82)
            colorplt = 'brown'
            print(1)
        elif label[i] == 'Dog':
            # colorplt = plt.cm.Set1(1.64)
            colorplt = 'red'
            print(2)
        # plt.text(data[i, 0], data[i, 1], str(label[i]), color=plt.cm.Set1(label[i] / 10),
        #          fontdict={'weight': 'bold', 'size': 7})
        # plt.text(data[i, 0], data[i, 1], str(label[i]), color=colorplt,
        #          fontdict={'weight': 'bold', 'size': 7})
        plt.scatter(data[i, 0], data[i, 1], color=colorplt, marker='o', s=50)
    plt.xticks()		# 指定坐标的刻度
    plt.yticks()
    plt.title(title, fontsize=14)
    # 返回值
    return fig

# 主函数，执行t-SNE降维
def main():
    data, label , n_samples, n_features = get_data()		# 调用函数，获取数据集信息

    my_allvector, my_alllabel = get_mydata()

    print('Starting compute t-SNE Embedding...')
    # print(label.shape)
    ts = TSNE(n_components=2, init='pca', random_state=0)
    # t-SNE降维
    reslut = ts.fit_transform(my_allvector)
    # 调用函数，绘制图像
    fig = plot_embedding(reslut, my_alllabel, 't-SNE Embedding of digits')
    # 显示图像
    plt.show()


# 主函数
if __name__ == '__main__':
    main

get_data()

加载MINIST数据集，来源网络

def get_data():
    """
    :return: 数据集、标签、样本数量、特征数量
    """
    digits = datasets.load_digits(n_class=10)
    data = digits.data		# 图片特征
    label = digits.target		# 图片标签
    n_samples, n_features = data.shape		# 数据集的形状
    return data, label, n_samples, n_features

get_mydata()

加载自己的数据集
class_name为保存每类数据的文件名，文件中每一行代表一个样本
class_label为标签

def get_mydata():
    class_name = ['CatVector.txt', 'DogVector.txt']
    class_label = ['Cat', 'Dog']
    for i in range(len(class_name)):
        vector = np.loadtxt(PATH + class_name[i])
        # vector = np.expand_dims(vector, axis = 0)
        for j in range(vector.shape[0]):
            my_label.append(class_label[i])
        if i == 0:
            allvector = vector
        else:
            allvector = np.concatenate((allvector, vector), axis=0)

    print(allvector.shape)
    print(np.array(my_label).shape)

    return allvector, np.array(my_label)

TSNE()参数

这个博主整理的很好
链接：link

matplotlib.pyplot.scatter()中颜色问题

颜色对照表：
链接：link