tsne降维python可视化

import time import numpy as np import pandas as pd import matplotlib.pyplot as plt from tensorflow.python.keras.utils import np_utils from sklearn.manifold import TSNE from sklearn.preprocessing import LabelEncoder df = pd.read_excel(r'C:/Users/1/Desktop/TSNE/data.xlsx', header=None) # 载入数据(4500, 2049),最后一列为标签 X = np.expand_dims(df.values[:, 0:2048].astype(float), axis=2) # (4500, 2048, 1) float64 # print(X.shape, X.dtype) Y = df.values[:, 2048] # (4500,) float64 # print(Y.shape, Y.dtype) # time.sleep(10) encoder = LabelEncoder() # 把标签做成one_hot encoding的形式 Y_encoded = encoder.fit_transform(Y) Y_onehot = np_utils.to_categorical(Y_encoded) # (4500, 10) float32 # print(Y_onehot.shape, Y_onehot.dtype) # time.sleep(10) X = X.reshape(4500, 2048) # 根据文件大小自行调试，4500是行数，2048是列数，此列数比文件中少一列 # print(X.shape, X.dtype) # 加载数据 def get_data(): """ :return: 数据集、标签、样本数量、特征数量 """ data = X label = Y n_samples = 4500 # 对应reshape中的行数 n_features = 2048 # 对应reshape中的列数 return data, label, n_samples, n_features # 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(figsize=(5.5, 4)) # Set the figure size in inches figsize=(7.3/2.54, 4.7/2.54) ax = plt.subplot(111) # 创建子图，经过验证111正合适，尽量不要修改 # 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] == 0: plt.plot(data[i, 0], data[i, 1], 'o', markerfacecolor='#C25c63', markeredgecolor='#C25c63', markersize=4) elif label[i] == 1: plt.plot(data[i, 0], data[i, 1], 'o', markerfacecolor='#357fae', markeredgecolor='#357fae', markersize=4) elif label[i] == 2: plt.plot(data[i, 0], data[i, 1], 'o', markerfacecolor='#FFDC91CC', markeredgecolor='#FFDC91CC', markersize=4) elif label[i] == 3: plt.plot(data[i, 0], data[i, 1], 'o', markerfacecolor='#FFC0CB', markeredgecolor='#FFC0CB', markersize=4) elif label[i] == 4: plt.plot(data[i, 0], data[i, 1], 'o', markerfacecolor='#FF00FF', markeredgecolor='#FF00FF', markersize=4) elif label[i] == 5: plt.plot(data[i, 0], data[i, 1], 'o', markerfacecolor='#87CEFA', markeredgecolor='#87CEFA', markersize=4) elif label[i] == 6: plt.plot(data[i, 0], data[i, 1], 'o', markerfacecolor='#FAEBD7', markeredgecolor='#FAEBD7', markersize=4) elif label[i] == 7: plt.plot(data[i, 0], data[i, 1], 'o', markerfacecolor='#98FB98', markeredgecolor='#98FB98', markersize=4) elif label[i] == 8: plt.plot(data[i, 0], data[i, 1], 'o', markerfacecolor='#FFA500', markeredgecolor='#FFA500', markersize=4) else: plt.plot(data[i, 0], data[i, 1], 'o', markerfacecolor='#7876B1FF', markeredgecolor='#7876B1FF', markersize=4) plt.xlim() plt.ylim() plt.title(title, fontsize=15, x=0.5, y=1) 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 print('Starting compute t-SNE Embedding...') ts = TSNE(n_components=2, init='pca', random_state=0) reslut = ts.fit_transform(data) # t-SNE降维 fig = plot_embedding(reslut, label, 'Input data visualisation') # 调用函数，绘制图像 plt.savefig("TSNE.png", dpi=600, format="png") # 显示图像 plt.show() # 主函数 if __name__ == '__main__': main()