AGNES¶

def fit(self, dataset: list[list[float]]) -> AGNES:
    """Fit the model."""
    self._validate(dataset)
    self.linkage_matrix_ = []
    n = len(dataset)
    self.clusters_ = [ClusterTree(cluster_index=i, sample_indices=[i]) for i in range(n)]
    self._cluster_index = n
    self.distance_matrix_ = self._get_init_distance_matrix(dataset)
    while len(self.clusters_) > self.n_cluster:
        (i, j), cluster_ij_distance = self._get_closest_clusters()
        # merge cluster_i and cluster_j
        self._merge_clusters(i, j, cluster_ij_distance)
        # update distance matrix
        self._update_distance_matrix(dataset, i, j)
    # build cluster_tree_
    self.cluster_tree_ = self._build_cluster_tree(n)
    # assign dataset labels
    self._get_labels(len(dataset))
    return self

def fit_predict(self, dataset: list[list[float]]) -> list[int]:
    """Fit the model and return the labels of each sample."""
    self.fit(dataset)
    return self.labels_.copy()

def plot_dendrogram(
    self,
    figure_name: str = "agnes_dendrogram.png",
    show: bool = False,
) -> None:
    """Plot the dendrogram of the clustering result.

    This method visualizes the hierarchical structure of the clustering
    using a dendrogram. It requires the number of clusters to be set to 1
    during initialization.

    Args:
        figure_name: The filename for saving the plot.
                           Defaults to "agnes_dendrogram.png".
        show: If True, displays the plot. Defaults to False.

    Raises:
        ValueError: If the number of clusters is not 1.

    Note:
        This method requires matplotlib and scipy to be installed.
    """
    if self.n_cluster != 1:
        msg = "The number of clusters should be 1 to plot dendrogram"
        raise ValueError(msg)
    # Plot the dendrogram
    plt.figure(figsize=(10, 7))
    dendrogram(np.array(self.linkage_matrix_))
    plt.title("AGNES Dendrogram")
    plt.xlabel("Sample Index")
    plt.ylabel("Distance")
    plt.savefig(f"{figure_name}", dpi=300, bbox_inches="tight")
    if show:
        plt.show()