I have a dataset comprise of 6654x1 rows. I have performed Expected Maximization clustering upon it using this code {https://www.mathworks.com/matlabcentral/fileexchange/8636-em_gm}
but my output clusters are not well defined. I have set k=4 cluster, and it shows me this.
Kindly guide me how do I proceed.
I can't understand your chart. If you have one feature for each row, what is this two-dimensional chart? What are the axis?
Note: E-M algorithm is a technique used in a broad class of models. What you have been doing is Gaussian Mixture Modelling.
Oh I think you have used the wrong model. There are two types of clustering.
1. Hard clustering (Clusters are not overlapping)
2. Soft clustering (Clusters are overlapping)
Gaussian Mixture Modelling is a soft clustering model. You should use hard clustering models (Like k-Means clustering) to get crisp clusters.
- Badges
- Science topic