MNIST Data plot in interactive 3d for ipython notebook

scatter-3d-tensorflow-mnist-datas.py

from tensorflow.keras.datasets import mnist
from sklearn.manifold import TSNE

import pandas as pd
import plotly.express as px

#load and adjust mnist data | Mnistのデータをロード
(_, _), (x_test, y_test) = mnist.load_data()
_, x_test = np.array(x_train, np.float32), np.array(x_test, np.float32)
_, x_test = x_train / 255., x_test / 255.
x_test = x_test.reshape(x_test.shape[0], x_test.shape[1]*x_test.shape[2])
#Cut to 1000 , cause too slow to process all | 多すぎると遅いので1000個へ
x_test = x_test[:1000]
y_test = y_test[:1000]

#convert 784 -> 3 dimention | 784次元から3次元に次元圧縮(3次元表示のため)
x_test_transformed_3 = TSNE(n_components=3).fit_transform(x_test)

#Show 3D Plotting | ここで3次元のプロット
df=pd.DataFrame(x_test_transformed_3,columns=list("XYZ"))
df["label"]=np.array(y_test)
fig = px.scatter_3d(df, x='X', y='Y', z='Z',  color='label', size_max=4, opacity=0.7)
fig.show()