Mean-shift-cluster

A tool, mean-shift-cluster is developed, which accepts both circular-linear data and linear data. Mean-shift-cluster supports three kernels: flat kernel, Gaussian kernel and truncated Gaussian kernel.

Usage

Mean-shift-cluster provides three APIs:

• fit: Fit input dataset using mean shift algorithm. Cluster centers and labels for each sample can be accessed.
• get_cluster_info: Provide basic information of each cluster(mode), using input dataset.
• predict: Predict the cluster results of any dataset, using the fitted mean-shift-cluster instance.

The inputs are:

• kernel: Type is string. Options are: flat, gaussian, truncated_gaussian
• kernel_parameters: type is List[float].
  • For flat kernel: [bandwidth]
  • For Gaussian kernel: [beta]
  • For truncated Gaussian kernel: [bandwidth, beta]
• data_type: Type is string. Options are: circular-linear, linear

To run example.py from this directory:

cd mean-shift
poetry run python example.py

A simple usage example is in example.py. And more details are in the report.

The script used when writing the report is reporter.py.