如何绘制 scikit 的 t-sne 输出数组？
Posted
技术标签:
【中文标题】如何绘制 scikit 的 t-sne 输出数组？【英文标题】：How to plot scikit's t-sne output array? 【发布时间】：2017-12-24 19:16:21 【问题描述】：我有一个简单的问题：我想绘制 scikit 的TSNE 的结果。这是我的工作：
import pandas as pd
import matplotlib.pyplot as plt
from sklearn.manifold import TSNE
x = df
model = TSNE(n_components=2, random_state=0)
model.fit_transform(x)
也就是说，我有一个相当大的数据框df。我运行TSNE。结果是一个数组，格式为
array([[ 76.29642863, -37.2316455 ],
[ 83.28206784, -35.19983576],
[ -10.93624436, -109.37550438],
[ -3.39073195, -107.29288942],
[ 67.66224215, -2.2439687 ],
[ 30.82205017, -19.28584524],
[ 47.35280231, -27.03922442],
[ 81.63500728, -14.94290102],
[ 36.48278883, -76.13490133],
[ 51.12458861, -12.63904649],
[ 27.37677179, -63.35607228],
[ 57.70966133, -22.75333134],
[ -60.32088017, 76.30353357],
[ 33.80070371, -5.59486463],
[ 27.80748222, -43.02255416],
[ 8.96419645, -5.12084324],
[ 100.61407581, -50.03230332],
[ 41.32789364, 73.4504622 ],
[ 20.51559019, -2.0397622 ],
[ 8.72757049, 9.42719579],
[ -97.09646231, 6.75980242],
[ -76.37504347, -67.92449062],
[ 40.8922133 , -54.64473336],
[ 21.07407628, 88.54593568],
[ 46.75201811, -43.85971949],
[ -62.77960662, -73.14171491],
[ 83.01345537, -52.70083606],
[ 84.660757 , 31.84526227],
[ -45.30560344, -80.44506703],
[ 11.42866734, -121.98241383],
[ 45.33793383, 11.43223043],
[ 48.66860826, 22.27729015],
[ -36.32341645, 89.51767685],
[ 48.29651789, 40.73176577],
[ 6.83643714, 70.83466828],
[ 31.13233995, 63.47771401],
[ 34.78917733, 22.13881353],
[ 43.99847623, 31.40394748],
[ -76.43334995, -46.1576235 ],
[ 16.35274296, 24.83566003],
[ 63.76553398, 43.75524517],
[ 17.8211855 , 66.45561224],
[ 24.03955628, 33.26653325],
[ 34.70078871, 42.87820685],
[ 55.27398051, 63.73362412],
[ 59.86821001, 19.59977048],
[ 3.18558424, 51.5108358 ],
[ 21.17605571, 44.4258268 ],
[ 70.05979974, 58.17860968],
[ 13.11104526, 51.09243066],
[ 4.38454865, 40.35813416],
[ -46.6602056 , 32.57412335],
[ 64.7773806 , -58.23294757],
[ -9.77563246, 22.91711485],
[ -4.63795974, 34.74267219],
[ -21.0280986 , 42.4962899 ],
[ -15.06925112, 33.66875649],
[ -23.38102294, 52.59088233],
[ -11.50369006, 56.07444623],
[ -23.69948462, -81.30270807],
[ -27.78705573, 62.36384916],
[ -36.11364478, 38.5400532 ],
[ -34.22060036, 47.86420995],
[-130.95539878, 7.47698021],
[ -99.13256617, -10.0366437 ],
[ -59.63287577, 31.69736952],
[ -5.51708328, -68.92901655],
[ -64.35945061, -0.92522887],
[ -42.98021829, 1.35631152],
[ -51.40872526, 45.27740173],
[ -53.20299633, 3.93563808],
[ -33.41025132, 0.61349718],
[ -64.20144157, 9.83423394],
[ -43.76759426, -11.21050429],
[ 2.95926436, -27.24926603],
[ 4.92291235, -40.27121135],
[ -32.32069564, -14.38969796],
[ -21.41289045, -8.75338496],
[ 42.71465077, 56.97575024],
[ -77.00615794, 36.85072343],
[ -22.37961976, -60.39178939],
[ -2.91171663, -50.77959613],
[ -50.60028744, -22.71077905],
[ -80.14729221, 9.56335332],
[ -62.70963932, -22.9475924 ],
[ -62.51993937, -36.2723845 ],
[ -71.86583485, 21.0174738 ],
[ -38.72063372, -27.05935043],
[ -47.74223044, -34.38562981],
[ 53.9395216 , 51.75186133],
[ -23.80825858, -43.32942045],
[ -16.85950018, -50.83328698],
[ -9.79065232, -29.92011604],
[ -19.76104305, -33.2619343 ],
[ -16.9120824 , -21.3528993 ],
[ 65.35647493, -13.17005252],
[ 42.60852749, -17.24403937],
[ 30.16514905, 53.77173972],
[ 21.99950701, 10.54130299],
[ 44.00495851, 88.05582053],
[ 82.9521465 , 14.17920421],
[ 6.0245289 , 86.73697853],
[ 97.38831046, 59.19700621],
[ 70.64054431, 29.54171715],
[ 57.39849223, 32.87309563],
[ 34.04147517, 32.74829322],
[ 12.81861228, 36.26916722],
[ -10.63982828, 44.92478561],
[ -26.58140799, 31.14874014],
[ -43.6592727 , 57.07683942],
[ -39.30343574, 23.39181145],
[ -55.93416401, 17.19611905],
[ -45.80009451, 12.10167266],
[ -54.36943413, -6.95559746],
[ 26.28194642, -30.60665791],
[ -63.04425748, 51.18032057],
[ -77.67041924, -1.10858794],
[ -5.38753713, -15.08877141],
[ -85.79812772, 23.3840777 ],
[ -42.34964347, -42.59061027],
[ -32.65764819, -47.82663035],
[ -27.87113561, -26.01927025],
[ -28.94307267, 10.93519372],
[ -35.42888905, -58.36504075],
[ -21.44710197, 86.52932663],
[ -11.41769981, -40.65429075],
[ -47.33535922, -53.86342049],
[ -31.07016357, -36.23505085]])
我现在正在尝试这样绘制数组：
plt.scatter(x[50:,] , x[:,50])
但是，我抛出了错误unhashable type: 'slice'。我做错了什么？
【问题讨论】：
你没有使用fit_transform()返回的TSNE值。它不会就地转换变量x。
【参考方案1】：
x 仍然是一个数据框，您应该将拟合的输出保存到变量中
res = model.fit_transform(x)
plt.scatter(res[50:,0] , res[50:,1])
然后你可以绘制 numpy 数组而不是数据框
【讨论】：
酷，我知道这是怎么回事！但是，我遇到了另一个错误：IndexError: index 50 is out of bounds for axis 1 with size 2- 我想我切错了。抱歉新手问题:(
我解决了！我只是将数组转换为熊猫数据框。 pd_res = pd.DataFrame(res[0:, 0:,]) 谢谢提醒！以上是关于如何绘制 scikit 的 t-sne 输出数组？的主要内容，如果未能解决你的问题，请参考以下文章
IndexError：使用 scikit-learn 绘制 ROC 曲线时数组索引过多？
如何将 Scikit Learn OneVsRestClassifier 预测方法输出转换为谷歌云 ML 的密集数组？