t-sne visualization in python, with tooltips on points

t-sne.py

import numpy as np
from sklearn.manifold import TSNE
import plotly.express as px
import pandas as pd
import glob
import json

# Sample data - replace with your own dataset
# data = np.random.rand(100, 20)  # 100 samples, 20 features

matrix = []
labels = []
for i in glob.glob("cache/embedding.specter_v2/*json"):
  data = json.load(open(i))
  if not data or not 'embedding' in data or not data['embedding'] or not data['embedding']['vector']:
    continue
  matrix.append(data['embedding']['vector'])
  labels.append(data['title'])

# create panda from matri
data = np.array(matrix)

# labels =   # Sample labels

# Perform t-SNE
tsne = TSNE(n_components=2, random_state=42)
tsne_results = tsne.fit_transform(data)

# Create a DataFrame for plotting
df = pd.DataFrame({
    'x': tsne_results[:, 0],
    'y': tsne_results[:, 1],
    'label': labels,
})


# Create interactive plot with tooltips
fig = px.scatter(
    df, x='x', y='y', 
    hover_data=['label'],
    title='t-SNE Visualization'
)

fig.update_traces(marker=dict(size=10))
fig.update_layout(
    hoverlabel=dict(
        bgcolor="white",
        font_size=12
    )
)

fig.show()