Print out the text value of the points on a cluster when using UMAP and HDBScan and BERT sentence transformer - bert-language-model

Question

I have seen a number of questions similar to this but my cluster labels consist of sentence embeddings, thus a better question may be how do I get text values from the sentence embeddings?
How can I get from my sentence embeddings to print a text output?
umap_embeddings = umap.UMAP(n_neighbors=50,
n_components=5,
metric='cosine').fit_transform(embeddings)
cluster = hdbscan.HDBSCAN(min_cluster_size=3,
metric='euclidean',
cluster_selection_method='eom').fit(umap_embeddings)
# Prepare data
umap_data = umap.UMAP(n_neighbors=15, n_components=2, min_dist=0.0, metric='cosine', random_state=24).fit_transform(embeddings)
result = pd.DataFrame(umap_data, columns=['x', 'y'])
result['labels'] = cluster.labels_
# Visualize clusters
fig, ax = plt.subplots(figsize=(20, 10))
outliers = result.loc[result.labels == -1, :]
clustered = result.loc[result.labels != -1, :]
plt.scatter(outliers.x, outliers.y, color='#202020', s=25)
plt.scatter(clustered.x, clustered.y, c=clustered.labels, s=25, cmap='hsv_r'
)
some previous answers have suggested;
textdata_with_label_113 = textData[clusterer.labels_ == 113]
However, this returns the embedded value oppoesd to the text value.

Answer 1

With more time on the problem I realised that the embeddings are in the same sequence as the original DF.
therefore you can work back quite easily.
lbls=[]
#seperate the clustered labels into seperate lists (0,1,2,3)
for x in range(len(clustered.labels)):
lbls.append(clustered[clustered.labels == x])
df_desc=[]
# extract the rows from the data frame using the lbls list and use column 6 only in my case
for x in range(len(lbls)):
df_desc.append(df.iloc[lbls[x].index,5])
for i in range(4):
txt = "Cluster {number}"
print(txt.format(number = i))
print(df_desc[i])