I've got the result shown in below figure for my clustering.
Are there any libraries which are similar to fvid_clusters which could produce the plots as shown below ? (USING PYTHON)
-
did you try matplotlibAkhilesh_IN– Akhilesh_IN2019-03-21 17:44:01 +00:00Commented Mar 21, 2019 at 17:44
-
Which vectorizer did you use to convert words to vectors ?mujjiga– mujjiga2019-03-21 18:24:06 +00:00Commented Mar 21, 2019 at 18:24
-
@mujjiga I use TfidfVectorizerTuck Yew– Tuck Yew2019-03-22 02:24:24 +00:00Commented Mar 22, 2019 at 2:24
-
@TuckYew did you got any way to create diagram as above?Rehan– Rehan2019-11-23 20:44:05 +00:00Commented Nov 23, 2019 at 20:44
-
Did you solve the problem? @Rehanuser11603027– user116030272020-05-14 11:53:28 +00:00Commented May 14, 2020 at 11:53
|
Show 1 more comment
2 Answers
Plots are not exactly like you show, but there two tools that I like to use in order to explore the result of text clustering (both have binds for Python).
Termite: http://vis.stanford.edu/papers/termite
UMAP: https://umap-learn.readthedocs.io/en/latest/
Here is the result of a topic modeling performed with NMF (from my thesis) using the BBC Sports dataset.
1 Comment
user11603027
can you please share the code? Thank you in advance.
- Step 1: Vectorize the words into vectors (each of size n)
- Step 2: Use kmean to cluster them into k clusters
- Step 3: Run PCA on vectors produced in step 1 and reduce them to 2 dimensions (say x,y)
- Step 4: Plot the words at the location (x, y) where (x,y) are the dimensions reduced by the PCA from its full size vector of size n. Color the word corresponding to which cluster it falls identified in step 2 (use one color per cluster).