王者荣耀喜欢玩的英雄赢不了,不喜欢的却能赢?那你对比过他们的属性吗？

深碍√TFBOYSˉ_ 2022-09-06 15:27 80阅读 0赞

# 导入库 #

import pandas as pd
    import matplotlib.pyplot as plt
    import seaborn as sns
    from sklearn.mixture import GaussianMixture
    from sklearn.preprocessing import StandardScaler
    from pyecharts.charts import Pie
    from pyecharts import options as opts
    from collections import Counter

### [数据集点这里][Link 1] ###

# 1. 所有英雄属性 #

plt.rcParams['font.sans-serif'] = ['SimHei']  # 正常显示中文标签
    plt.rcParams['axes.unicode_minus'] = False  # 正常显示负号
    pd.set_option('display.max_columns', None)
    pd.set_option('display.max_rows', None)
    data = pd.read_csv('/home/mw/input/hero3741/all_hero_init_attr.csv', encoding='gb18030')

feature = ['英雄名字', '生存能力', '攻击伤害', '技能效果',
               '上手难度', '最大生命', '最大法力', '物理攻击',
               '法术攻击', '物理防御', '物理减伤率', '法术防御',
               '法术减伤率', '移速', '物理护甲穿透', '法术护甲穿透',
               '攻速加成', '暴击几率', '暴击效果', '物理吸血', '法术吸血',
               '冷却缩减', '攻击范围', '韧性', '生命回复', '法力回复']
    data_init = data[feature]
    data_init.head()

![watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L20wXzQ4NDA1Nzgx_size_16_color_FFFFFF_t_70][]

# 把空值设置为0
    data_init = data_init.fillna(0)
    # print(data_init.isnull().sum())
    ''' 正则匹配包含%%的数据 p = r'(%%)' a = data_init['物理减伤率'].astype("object").str.extract(p) print(a) '''
    
    features_remain = ['生存能力', '攻击伤害', '技能效果',
                        '上手难度', '最大生命', '最大法力', '物理攻击',
                        '法术攻击', '物理防御', '物理减伤率', '移速', '攻击范围', '生命回复', '法力回复']
    
    data_init = data_init[features_remain]
    
    data_init['物理减伤率'] = data_init['物理减伤率'].apply(lambda x: float(x.strip('%'))/100)
    
    data_init['攻击范围'] = data_init['攻击范围'].map({ '远程': 1, '近程': 0})
    
    # print(data_init)
    
    ss = StandardScaler()
    data_init = ss.fit_transform(data_init)
    
    gmm = GaussianMixture(n_components=20, covariance_type='full')
    gmm.fit(data_init)
    
    prediction = gmm.predict(data_init)
    # print(prediction)
    data.insert(0, '分组', prediction)
    data.to_csv('/home/mw/input/hero3741/all_hero_init_attr_our.csv', index=False, sep=',', encoding='gb18030')
    
    
    df = data[['分组', '英雄名字']]  # 获取需要的两列
    
    grouped = df.groupby(['分组'])  # 以”分组“列来进行分组
    k = []
    
    # 获取分组后的 组和值，保存为字典，放到列表中
    for name, group in grouped:
        k.append({ name: list(group['英雄名字'].values)})
    
    kk = []
    for i in k:
        for k, v in i.items():
            kk.append(v)
    
    length = []
    key = []
    for i in kk:
        key.append(str(i))
        length.append(len(i))

key, length

![watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L20wXzQ4NDA1Nzgx_size_16_color_FFFFFF_t_70 1][]

pie = Pie()
    pie.add("", [list(z) for z in zip(key, length)],
           radius=["30%", "75%"], center=["45%", "50%"], rosetype="radius",
           )
    pie.set_series_opts(label_opts=opts.LabelOpts(formatter="{b}, {d}%"))
    pie.set_global_opts(
                title_opts=opts.TitleOpts(title="英雄初始属性分类图"),
                legend_opts=opts.LegendOpts(
                    type_="scroll", pos_left="80%", orient="vertical"
                ),
            )
    pie.render_notebook()

![watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L20wXzQ4NDA1Nzgx_size_16_color_FFFFFF_t_70 2][]

# 2. 根据英雄属性进行聚类分析 #

data.head()

![watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L20wXzQ4NDA1Nzgx_size_16_color_FFFFFF_t_70 3][]

data_init = data[feature]

data_init = data_init.fillna(0)

corr = data_init[feature].corr()
    plt.figure(figsize=(14, 14))
    sns.heatmap(corr, annot=True)
    plt.show()

![watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L20wXzQ4NDA1Nzgx_size_16_color_FFFFFF_t_70 4][]

features_remain = ['生存能力', '攻击伤害', '技能效果',
                        '上手难度', '最大生命', '最大法力', '物理攻击',
                        '法术攻击', '物理防御', '物理减伤率', '移速', '攻击范围', '生命回复', '法力回复']

data_init = data_init[features_remain]
    data_init.head()

![watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L20wXzQ4NDA1Nzgx_size_16_color_FFFFFF_t_70 5][]

data_init['物理减伤率'] = data_init['物理减伤率'].apply(lambda x: float(x.strip('%'))/100)
    
    data_init['攻击范围'] = data_init['攻击范围'].map({ '远程': 1, '近程': 0})

data_init.head()

![watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L20wXzQ4NDA1Nzgx_size_16_color_FFFFFF_t_70 6][]

ss = StandardScaler()
    data_init = ss.fit_transform(data_init)

gmm = GaussianMixture(n_components=20, covariance_type='full')
    gmm.fit(data_init)

GaussianMixture(covariance_type='full', init_params='kmeans', max_iter=100,
                    means_init=None, n_components=20, n_init=1,
                    precisions_init=None, random_state=None, reg_covar=1e-06,
                    tol=0.001, verbose=0, verbose_interval=10, warm_start=False,
                    weights_init=None)

data.insert(0, '分组', prediction)
    data.to_csv('/home/mw/input/hero3741/all_hero_init_attr_our.csv', index=False, sep=',', encoding='gb18030')

![watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L20wXzQ4NDA1Nzgx_size_16_color_FFFFFF_t_70 7][]

pie = Pie()
    pie.add("", [list(z) for z in zip(key, length)],
           radius=["30%", "75%"],
                center=["40%", "50%"],
                rosetype="radius")
    pie.set_global_opts(
                title_opts=opts.TitleOpts(title="英雄聚类分布"),
                legend_opts=opts.LegendOpts(
                    type_="scroll", pos_left="80%", orient="vertical"
                ),
            )
    pie.render_notebook()

![watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L20wXzQ4NDA1Nzgx_size_16_color_FFFFFF_t_70 8][]

[Link 1]: https://jq.qq.com/?_wv=1027&k=vT8UXWt1
[watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L20wXzQ4NDA1Nzgx_size_16_color_FFFFFF_t_70]: /images/20220829/b0bfe7b151464830ad7d1d4986a87d61.png
[watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L20wXzQ4NDA1Nzgx_size_16_color_FFFFFF_t_70 1]: /images/20220829/9a39bd9c869b4153af6e90b1cf114a1f.png
[watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L20wXzQ4NDA1Nzgx_size_16_color_FFFFFF_t_70 2]: /images/20220829/481004332c854770b94a50a98139a90d.png
[watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L20wXzQ4NDA1Nzgx_size_16_color_FFFFFF_t_70 3]: /images/20220829/7b3cde6618864db184aa884e910f2fe8.png
[watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L20wXzQ4NDA1Nzgx_size_16_color_FFFFFF_t_70 4]: /images/20220829/3fccd9b4d01d4e7181026c23d630fe57.png
[watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L20wXzQ4NDA1Nzgx_size_16_color_FFFFFF_t_70 5]: /images/20220829/50cb4f4c733e4c98a67d7246299e49f3.png
[watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L20wXzQ4NDA1Nzgx_size_16_color_FFFFFF_t_70 6]: /images/20220829/a4ae34352418424bb25fc993ad8f1f8b.png
[watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L20wXzQ4NDA1Nzgx_size_16_color_FFFFFF_t_70 7]: /images/20220829/8a260d56f63d47de87578200f623cdc6.png
[watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L20wXzQ4NDA1Nzgx_size_16_color_FFFFFF_t_70 8]: /images/20220829/deb6a513773241fd99f91f18c59b10cc.png