LSA降维后层次聚类Dendrogram标注文本簇可读标签方法咨询

你之前绘制的树状图仅显示样本数量，核心原因是调用dendrogram函数时没有传入自定义labels参数，截断模式下会默认显示节点的样本计数。按照以下步骤修改即可实现和mtcars示例一致的自定义标签效果：

基础实现：直接使用样本名称作为标签 #

直接在调用绘图函数时传入labels参数即可，完整可运行代码如下：

import pandas as pd
from matplotlib import pyplot as plt
import numpy as np
from sklearn.feature_extraction.text import TfidfVectorizer
from sklearn.cluster import AgglomerativeClustering
from sklearn.decomposition import TruncatedSVD

# 加载IMDB数据集
df = pd.read_csv(
    'https://raw.githubusercontent.com/peetck/IMDB-Top1000-Movies/master/IMDB-Movie-Data.csv'
)

# 文本向量化 + LSA降维
vectorizer = TfidfVectorizer(stop_words='english', max_features=1000)
X = vectorizer.fit_transform(df.Description)
TSVD = TruncatedSVD(n_components=10)
X_reduced = TSVD.fit_transform(X)

# 训练层次聚类模型
model = AgglomerativeClustering(distance_threshold=0, n_clusters=None)
model = model.fit(X_reduced)

def plot_dendrogram(model, **kwargs):
    # 构建聚类链接矩阵
    counts = np.zeros(model.children_.shape[0])
    n_samples = len(model.labels_)
    for i, merge in enumerate(model.children_):
        current_count = 0
        for child_idx in merge:
            if child_idx < n_samples:
                current_count += 1  # 叶节点计数
            else:
                current_count += counts[child_idx - n_samples]
        counts[i] = current_count
    linkage_matrix = np.column_stack([model.children_, model.distances_, counts]).astype(float)
    return dendrogram(linkage_matrix, **kwargs)

# 核心修改：传入labels参数指定叶节点显示的文本，此处用电影名作为标签
out = plot_dendrogram(model, 
                p=3, 
                truncate_mode='level', 
                leaf_rotation=90,
                leaf_font_size=7,
                labels=df.Title.values)

# 调整布局避免标签被截断
plt.tight_layout()
plt.show()

进阶实现：为聚合节点生成语义标签 #

如果开启了层级截断，合并的非叶节点可以自定义生成语义标签，比如提取簇内文本的高频关键词作为标签，补充代码如下：

# 提取每个节点对应的样本索引映射
n_samples = len(model.labels_)
node_sample_map = {}
# 初始化叶节点的样本索引
for i in range(n_samples):
    node_sample_map[i] = [i]
# 递归生成聚合节点的样本索引
for i, merge in enumerate(model.children_):
    node_id = n_samples + i
    node_sample_map[node_id] = node_sample_map[merge[0]] + node_sample_map[merge[1]]

# 生成簇关键词的工具函数
def get_cluster_keywords(sample_indices, top_k=2):
    cluster_tfidf = X[sample_indices]
    mean_tfidf = np.array(cluster_tfidf.mean(axis=0)).flatten()
    top_word_ids = mean_tfidf.argsort()[-top_k:][::-1]
    return "/".join(vectorizer.get_feature_names_out()[top_word_ids])

# 重新生成所有显示节点的标签
ax = plt.gca()
final_labels = []
for leaf_id in out['leaves']:
    if leaf_id < n_samples:
        # 叶节点显示电影名
        final_labels.append(df.Title.iloc[leaf_id])
    else:
        # 聚合节点显示高频关键词
        final_labels.append(get_cluster_keywords(node_sample_map[leaf_id]))

# 替换x轴标签
ax.set_xticklabels(final_labels, rotation=90, fontsize=6)
plt.tight_layout()
plt.show()

内容的提问来源于stack exchange，提问作者osint_alex