Skip to content

module calkmeans

function calkmeans

1
2
3
4
5
6
7
8
9
calkmeans(
    image: DatasetReader,
    k: Optional[int] = None,
    algo: Tuple[str, ] = ('lloyd', 'elkan'),
    max_iter: int = 300,
    n_iter: int = 10,
    nodata: float = -99999.0,
    **kwargs
) → Dict[str, List[float]]

Calibrating KMeans Clustering Algorithm

This function calibrates the KMeans algorithm for satellite image classification. It can either find the optimal number of clusters (k) by evaluating the inertia over a range of cluster numbers or determine the best algorithm ('lloyd' or 'elkan') for a given k.

Parameters:

  • image (rasterio.io.DatasetReader): Optical image with 3D data.
  • k (Optional[int]): The number of clusters. If None, the function finds the optimal k.
  • algo (Tuple[str, ...]): Algorithms to evaluate ('lloyd', 'elkan').
  • max_iter (int): Maximum iterations for KMeans (default 300).
  • n_iter (int): Number of iterations or clusters to evaluate (default 10).
  • nodata (float): The NoData value to identify and handle in the data.
  • **kwargs: Additional arguments passed to sklearn.cluster.KMeans.

Returns:

  • Dict[str, List[float]]: A dictionary with algorithm names as keys and lists of inertia values as values.

Notes:

  • If k is None, the function evaluates inertia for cluster numbers from 1 to n_iter. - If k is provided, the function runs KMeans n_iter times for each algorithm to evaluate their performance. - The function handles NoData values using fuzzy matching to account for floating-point precision.

This file was automatically generated via lazydocs.