# -*- coding: utf-8 -*-

"""
@Datetime: 2019/3/31
@Author: Zhang Yafei
"""
# https://www.cnblogs.com/lc1217/p/6908031.html
import functools
import time

import pandas as pd
import numpy as np
from sklearn.cluster import AffinityPropagation
from sklearn.datasets.samples_generator import make_blobs
from sklearn.metrics.pairwise import cosine_similarity
from sklearn.metrics import euclidean_distances, silhouette_score
from sklearn.preprocessing import StandardScaler


def timeit(fun):
    @functools.wraps(fun)
    def wrapper(*args, **kwargs):
        start_time = time.time()
        res = fun(*args, **kwargs)
        print('运行时间为%.6f' % (time.time() - start_time))
        return res

    return wrapper


def init_sample():
    """
    第一步：生成测试数据
        1.生成实际中心为centers的测试样本300个，
        2.Xn是包含150个(x,y)点的二维数组
        3.labels_true为其对应的真是类别标签
    """
    # 生成的测试数据的中心点
    centers = [[1, 1], [-1, -1], [1, -1]]
    # 生成数据
    X, label_true = make_blobs(n_samples=150, centers=centers, cluster_std=0.5, random_state=0)
    return X, label_true


def simi_matrix(Xn):
    simi = []
    for m in Xn:
        ##每个数字与所有数字的相似度列表，即矩阵中的一行
        temp = []
        for n in Xn:
             ##采用负的欧式距离计算相似度
            s = np.sqrt((m[0]-n[0])**2 + (m[1]-n[1])**2)
            temp.append(s)
        simi.append(temp)
    return np.around(np.array(simi), decimals=8)


@timeit
def main():
    """ 2个特征 """
    Xn, label_true = init_sample()
    eu_simi_matrix = euclidean_distances(X=Xn, Y=Xn)
    print(eu_simi_matrix)
    p = -50   ##3个中心
    # p = np.min(eu_simi_matrix)  ##9个中心，
    # p = np.median(simi)  ##13个中心
    ap = AffinityPropagation(damping=0.5, max_iter=500, convergence_iter=30, preference=p).fit(Xn)
    cluster_centers_indices = ap.cluster_centers_indices_

    print(ap.labels_)
    for idx in cluster_centers_indices:
        print(Xn[idx])


if __name__ == '__main__':
    # main()
    """ 多个特征 """
    # 1. 读取数据
    beer = pd.read_csv('data.txt', sep=' ')
    # 2. 读取特征X, 并标准化
    X = beer[beer.columns[beer.columns != 'name']].values
    X = StandardScaler().fit_transform(X)
    # 3. 相似矩阵
    cosine_simi_matrix = cosine_similarity(X)
    # eu_simi_matrix = euclidean_distances(X=X, Y=X)
    # print(eu_simi_matrix)
    # p = -10
    # p = np.min(cosine_simi_matrix)  # 11个中心，
    # p = np.median(cosine_simi_matrix)
    # print(p)
    # 4. AP聚类
    # 选择最优参数
    scores = {}
    for p in range(-20, -10):
        ap = AffinityPropagation(damping=0.5, max_iter=500, convergence_iter=30, preference=p).fit(X)
        labels = ap.labels_
        print(set(labels).__len__())
        score = silhouette_score(X, labels)
        scores[p] = score
    best_p = sorted(scores.items(), key=lambda x: x[1], reverse=True)[0]
    print(best_p)
    # 进行聚类
    ap = AffinityPropagation(damping=0.5, max_iter=500, convergence_iter=30, preference=best_p[0]).fit(X)
    print(set(ap.labels_).__len__())
    cluster_centers_indices = ap.cluster_centers_indices_
    for idx in cluster_centers_indices:
        print(X[idx])