python爬取電競《絕地求生》比賽數據集分析

　　　　　　python爬取電競《絕地求生》比賽數據集分析

一，選題背景

　　電子競技（Electronic Sports）是電子游戲比賽達到“競技”層面的體育項目。電子競技就是利用電子設備作為運動器械進行的、人與人之間的智力和體力結合的比拼。通過電子競技，可以鍛煉和提高參與者的思維能力、反應能力、四肢協調能力和意志力，培養團隊精神，並且職業電競對體力也有較高要求。電子競技也是一種職業，和棋藝等非電子游戲比賽類似，2003年11月18日，國家體育局正式批准，將電子競技列為第99個正式體育競賽項目。2008年，國家體育總局將電子競技改批為第78號正式體育競賽項目。2018年雅加達亞運會將電子競技納為表演項目。數據來源：20G絕地求生比賽數據集。

 

二，設計方案

1，爬蟲名稱：python爬取電競《絕地求生》比賽數據集分析

2，爬蟲爬取的內容與數據特征分析

　　主要分成兩部分，一部分是玩家比賽的統計數據，以agg_match_stats開頭，一部分是玩家被擊殺的數據，以kill_match_stats開頭本次分析選取其中的兩個數據集進行分析

3，設計方案：

1. 飛機嗡嗡地，我到底跳哪里比較安全？

2. 我是該苟着不動，還是應該出去猛干？

3. 是該單打獨斗還是跟隊友一起配合？

4. 毒來了我跑不過毒怎么辦啊？

5. 什么武器最有用？

6. 近戰適合使用什么武器，狙擊適合使用什么武器呢？

7. 最后的毒圈一般會在哪里呢？

 

三，結果特征分析

1，頁面的結構與特征分析

 

 

 

 

 

 四，程序設計

1，# 使用pandas讀取數據

import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
import seaborn as sns
%matplotlib inline
plt.rcParams['font.sans-serif'] = ['SimHei'] # 指定默認字體
plt.rcParams['axes.unicode_minus'] = False # 解決保存圖像是負號'-'顯示為方塊的問題

# 使用pandas讀取數據
agg1 = pd.read_csv('/Users/apple/Desktop/pubg/aggregate/agg_match_stats_1.csv')

 

2，# 探索數據結構並數據清洗

agg1.head()

 

 

 

 

# 總共有13844275行玩家數據，15列

agg1.shape

 

 

agg1.columns

 

 

 

 

 agg1.info()

 

 

 

 3，數據分析與可視化

# 丟棄重復數據

agg1.drop_duplicates(inplace=True)

agg1.loc[1]

 

 

 

 # 添加是否成功吃雞列

agg1['won'] = agg1['team_placement'] == 1

 

# 添加是否搭乘過車輛列

agg1['drove'] = agg1['player_dist_ride'] != 0

 

 

我是該苟着不動，還是應該出去猛干

agg1.loc[agg1['player_kills'] < 40, ['player_kills', 'won']].groupby('player_kills').won.mean().plot.bar(figsize=(15,6), rot=0)
plt.xlabel('擊殺人數', fontsize=14)
plt.ylabel("吃雞概率", fontsize=14)
plt.title('擊殺人數與吃雞概率的關系', fontsize=14)

 

 

 

 不同模式下的平均擊殺人數：

agg1.groupby('party_size').player_kills.mean()

 

 

 

 

 

 

 

g = sns.FacetGrid(agg1.loc[agg1['player_kills']<=10, ['party_size', 'player_kills']], row="party_size", size=4, aspect=2)
g = g.map(sns.countplot, "player_kills")



party_size=1

 

 

 

 party_size=2

 

 

 

  party_size=4

 

 

 

 是該單打獨斗還是跟隊友一起配合？

agg1.loc[agg1['party_size']!=1, ['player_assists', 'won']].groupby('player_assists').won.mean().plot.bar(figsize=(15,6), rot=0)
plt.xlabel('助攻次數', fontsize=14)
plt.ylabel("吃雞概率", fontsize=14)
plt.title('助攻次數與吃雞概率的關系', fontsize=14)

 

 

 

 “毒來了我跑不過毒怎么辦啊”之車輛到底有多重要？

agg1.groupby('drove').won.mean().plot.barh(figsize=(6,3))
plt.xlabel("吃雞概率", fontsize=14)
plt.ylabel("是否搭乘過車輛", fontsize=14)
plt.title('搭乘車輛與吃雞概率的關系', fontsize=14)
plt.yticks([1,0],['是','否'])

 

 

 

 

 

dist_ride = agg1.loc[agg1['player_dist_ride']<12000, ['player_dist_ride', 'won']]

labels=["0-1k", "1-2k", "2-3k", "3-4k","4-5k", "5-6k", "6-7k", "7-8k", "8-9k", "9-10k", "10-11k", "11-12k"]
dist_ride['drove_cut'] = pd.cut(dist_ride['player_dist_ride'], 12, labels=labels)

dist_ride.groupby('drove_cut').won.mean().plot.bar(rot=60, figsize=(8,4))
plt.xlabel("搭乘車輛里程", fontsize=14)
plt.ylabel("吃雞概率", fontsize=14)
plt.title('搭乘車輛里程與吃雞概率的關系', fontsize=14)

 

 

 

 

 match_unique = agg1.loc[agg1['party_size'] == 1, 'match_id'].unique()

match_unique = agg1.loc[agg1['party_size'] == 1, 'match_id'].unique()

 

把玩家被擊殺的數據導入進來

# 先把玩家被擊殺的數據導入進來並探索數據
death1 = pd.read_csv('/Users/apple/Desktop/pubg/deaths/kill_match_stats_final_1.csv')

death1.head()

 

 

 

 

death1.info()

death1.shape

death1_solo = death1[death1['match_id'].isin(match_unique)]

death1_solo.info()

 

 

 

 

 

 

 飛機嗡嗡地，我到底跳哪里比較安全

 

# 只統計單人模式，篩選存活不超過180秒的玩家數據
death_180_seconds_erg = death1_solo.loc[(death1_solo['map'] == 'ERANGEL')&(death1_solo['time'] < 180)&(death1_solo['victim_position_x']>0), :].dropna()
death_180_seconds_mrm = death1_solo.loc[(death1_solo['map'] == 'MIRAMAR')&(death1_solo['time'] < 180)&(death1_solo['victim_position_x']>0), :].dropna()

death_180_seconds_erg.shape

death_180_seconds_mrm.shape

 

 

 

 

# 選擇存活不過180秒的玩家死亡位置
data_erg = death_180_seconds_erg[['victim_position_x', 'victim_position_y']].values
data_mrm = death_180_seconds_mrm[['victim_position_x', 'victim_position_y']].values

# 重新scale玩家位置
data_erg = data_erg*4096/800000
data_mrm = data_mrm*1000/800000

from scipy.ndimage.filters import gaussian_filter
import matplotlib.cm as cm
from matplotlib.colors import Normalize
from scipy.misc.pilutil import imread

from scipy.ndimage.filters import gaussian_filter
import matplotlib.cm as cm
from matplotlib.colors import Normalize

def heatmap(x, y, s, bins=100):
heatmap, xedges, yedges = np.histogram2d(x, y, bins=bins)
heatmap = gaussian_filter(heatmap, sigma=s)

extent = [xedges[0], xedges[-1], yedges[0], yedges[-1]]
return heatmap.T, extent

bg = imread('/Users/apple/Desktop/pubg/erangel.jpg')
hmap, extent = heatmap(data_erg[:,0], data_erg[:,1], 4.5)
alphas = np.clip(Normalize(0, hmap.max(), clip=True)(hmap)*4.5, 0.0, 1.)
colors = Normalize(0, hmap.max(), clip=True)(hmap)
colors = cm.Reds(colors)
colors[..., -1] = alphas

fig, ax = plt.subplots(figsize=(24,24))
ax.set_xlim(0, 4096); ax.set_ylim(0, 4096)
ax.imshow(bg)
ax.imshow(colors, extent=extent, origin='lower', cmap=cm.Reds, alpha=0.9)
plt.gca().invert_yaxis()

 

 

 

 

bg = imread('/Users/apple/Desktop/pubg/miramar.jpg')
hmap, extent = heatmap(data_mrm[:,0], data_mrm[:,1], 4)
alphas = np.clip(Normalize(0, hmap.max(), clip=True)(hmap)*4, 0.0, 1.)
colors = Normalize(0, hmap.max(), clip=True)(hmap)
colors = cm.Reds(colors)
colors[..., -1] = alphas

fig, ax = plt.subplots(figsize=(24,24))
ax.set_xlim(0, 1000); ax.set_ylim(0, 1000)
ax.imshow(bg)
ax.imshow(colors, extent=extent, origin='lower', cmap=cm.Reds, alpha=0.9)
#plt.scatter(plot_data_mr[:,0], plot_data_mr[:,1])
plt.gca().invert_yaxis()

 

 

 

 

 最后的毒圈一般會在哪里呢？

這里選取每場比賽第一名和第二名的位置數據，因為第一名和第二名所在的位置基本上就是最后的毒圈所在的位置

death_final_circle_erg = death1_solo.loc[(death1_solo['map'] == 'ERANGEL')&(death1_solo['victim_placement'] == 2)&(death1_solo['victim_position_x']>0)&(death1_solo['killer_position_x']>0), :].dropna()
death_final_circle_mrm = death1_solo.loc[(death1_solo['map'] == 'MIRAMAR')&(death1_solo['victim_placement'] == 2)&(death1_solo['victim_position_x']>0)&(death1_solo['killer_position_x']>0), :].dropna()

print(death_final_circle_erg.shape)
print(death_final_circle_mrm.shape)

final_circle_erg = np.vstack([death_final_circle_erg[['victim_position_x', 'victim_position_y']].values,
death_final_circle_erg[['killer_position_x', 'killer_position_y']].values])*4096/800000
final_circle_mrm = np.vstack([death_final_circle_mrm[['victim_position_x', 'victim_position_y']].values,
death_final_circle_mrm[['killer_position_x', 'killer_position_y']].values])*1000/800000

 

 

 

 

bg = imread('/Users/apple/Desktop/pubg/erangel.jpg')
hmap, extent = heatmap(final_circle_erg[:,0], final_circle_erg[:,1], 1.5)
alphas = np.clip(Normalize(0, hmap.max(), clip=True)(hmap)*1.5, 0.0, 1.)
colors = Normalize(0, hmap.max(), clip=True)(hmap)
colors = cm.Reds(colors)
colors[..., -1] = alphas

fig, ax = plt.subplots(figsize=(24,24))
ax.set_xlim(0, 4096); ax.set_ylim(0, 4096)
ax.imshow(bg)
ax.imshow(colors, extent=extent, origin='lower', cmap=cm.Reds, alpha=0.9)
#plt.scatter(plot_data_er[:,0], plot_data_er[:,1])

plt.gca().invert_yaxis()

 

 

 

 

bg = imread('/Users/apple/Desktop/pubg/miramar.jpg')
hmap, extent = heatmap(final_circle_mrm[:,0], final_circle_mrm[:,1], 1.5)
alphas = np.clip(Normalize(0, hmap.max(), clip=True)(hmap)*1.5, 0.0, 1.)
colors = Normalize(0, hmap.max(), clip=True)(hmap)
colors = cm.Reds(colors)
colors[..., -1] = alphas

fig, ax = plt.subplots(figsize=(24,24))
ax.set_xlim(0, 1000); ax.set_ylim(0, 1000)
ax.imshow(bg)
ax.imshow(colors, extent=extent, origin='lower', cmap=cm.Reds, alpha=0.9)
#plt.scatter(plot_data_mr[:,0], plot_data_mr[:,1])
plt.gca().invert_yaxis()

 

 

 

 什么武器最有用？

erg_died_of = death1.loc[(death1['map']=='ERANGEL')&(death1['killer_position_x']>0)&(death1['victim_position_x']>0)&(death1['killed_by']!='Down and Out'),:]
mrm_died_of = death1.loc[(death1['map']=='MIRAMAR')&(death1['killer_position_x']>0)&(death1['victim_position_x']>0)&(death1['killed_by']!='Down and Out'),:]

print(erg_died_of.shape)
print(mrm_died_of.shape)

 

 

 

 

erg_died_of['killed_by'].value_counts()[:10].plot.barh(figsize=(10,5))
plt.xlabel("被擊殺人數", fontsize=14)
plt.ylabel("擊殺的武器", fontsize=14)
plt.title('武器跟擊殺人數的統計(絕地海島艾倫格)', fontsize=14)
plt.yticks(fontsize=12)

 

 

 

 

狙擊適合使用什么武器？

 

# 把位置信息轉換成距離，以“米”為單位
erg_distance = np.sqrt(((erg_died_of['killer_position_x']-erg_died_of['victim_position_x'])/100)**2 + ((erg_died_of['killer_position_y']-erg_died_of['victim_position_y'])/100)**2)



mrm_distance = np.sqrt(((mrm_died_of['killer_position_x']-mrm_died_of['victim_position_x'])/100)**2 + ((mrm_died_of['killer_position_y']-mrm_died_of['victim_position_y'])/100)**2)



sns.distplot(erg_distance.loc[erg_distance<400])

 

 

 

 

erg_died_of.loc[(erg_distance > 800)&(erg_distance < 1500), 'killed_by'].value_counts()[:10].plot.bar(rot=30)
plt.xlabel("狙擊的武器", fontsize=14)
plt.ylabel("被狙擊的人數", fontsize=14)
plt.title('狙擊武器跟擊殺人數的統計(絕地海島艾倫格)', fontsize=14)
plt.yticks(fontsize=12)

 

 

 

 

mrm_died_of.loc[(mrm_distance > 800)&(mrm_distance < 1000), 'killed_by'].value_counts()[:10].plot.bar(rot=30)
plt.xlabel("狙擊的武器", fontsize=14)
plt.ylabel("被狙擊的人數", fontsize=14)
plt.title('狙擊武器跟擊殺人數的統計(熱情沙漠米拉瑪)', fontsize=14)
plt.yticks(fontsize=12)

 

 

 

 近戰適合使用什么武器？

erg_died_of.loc[erg_distance<10, 'killed_by'].value_counts()[:10].plot.bar(rot=30)
plt.xlabel("近戰武器", fontsize=14)
plt.ylabel("被擊殺的人數", fontsize=14)
plt.title('近戰武器跟擊殺人數的統計(絕地海島艾倫格)', fontsize=14)
plt.yticks(fontsize=12)

 

 

 

 

mrm_died_of.loc[mrm_distance<10, 'killed_by'].value_counts()[:10].plot.bar(rot=30)
plt.xlabel("近戰武器", fontsize=14)
plt.ylabel("被擊殺的人數", fontsize=14)
plt.title('近戰武器武器跟擊殺人數的統計(熱情沙漠米拉瑪)', fontsize=14)
plt.yticks(fontsize=12)

 

 

 

 Top10武器在各距離下的擊殺百分比

erg_died_of['erg_dist'] = erg_distance
erg_died_of = erg_died_of.loc[erg_died_of['erg_dist']<800, :]
top_weapons_erg = list(erg_died_of['killed_by'].value_counts()[:10].index)
top_weapon_kills = erg_died_of[np.in1d(erg_died_of['killed_by'], top_weapons_erg)].copy()
top_weapon_kills['bin'] = pd.cut(top_weapon_kills['erg_dist'], np.arange(0, 800, 10), include_lowest=True, labels=False)
top_weapon_kills_wide = top_weapon_kills.groupby(['killed_by', 'bin']).size().unstack(fill_value=0).transpose()



top_weapon_kills_wide = top_weapon_kills_wide.div(top_weapon_kills_wide.sum(axis=1), axis=0)



from bokeh.models.tools import HoverTool
from bokeh.palettes import brewer
from bokeh.plotting import figure, show, output_notebook
from bokeh.models.sources import ColumnDataSource

def stacked(df):
df_top = df.cumsum(axis=1)
df_bottom = df_top.shift(axis=1).fillna(0)[::-1]
df_stack = pd.concat([df_bottom, df_top], ignore_index=True)
return df_stack

hover = HoverTool(
tooltips=[
("index", "$index"),
("weapon", "@weapon"),
("(x,y)", "($x, $y)")
],
point_policy='follow_mouse'
)

areas = stacked(top_weapon_kills_wide)

colors = brewer['Spectral'][areas.shape[1]]
x2 = np.hstack((top_weapon_kills_wide.index[::-1],
top_weapon_kills_wide.index)) /0.095

TOOLS="pan,wheel_zoom,box_zoom,reset,previewsave"
output_notebook()
p = figure(x_range=(1, 800), y_range=(0, 1), tools=[TOOLS, hover], plot_width=800)
p.grid.minor_grid_line_color = '#eeeeee'

source = ColumnDataSource(data={
'x': [x2] * areas.shape[1],
'y': [areas[c].values for c in areas],
'weapon': list(top_weapon_kills_wide.columns),
'color': colors
})

p.patches('x', 'y', source=source, legend="weapon",
color='color', alpha=0.8, line_color=None)
p.title.text = "Top10武器在各距離下的擊殺百分比（絕地海島艾倫格）"
p.xaxis.axis_label = "擊殺距離（0-800米）"
p.yaxis.axis_label = "百分比"

show(p)



mrm_died_of['erg_dist'] = mrm_distance
mrm_died_of = mrm_died_of.loc[mrm_died_of['erg_dist']<800, :]
top_weapons_erg = list(mrm_died_of['killed_by'].value_counts()[:10].index)
top_weapon_kills = mrm_died_of[np.in1d(mrm_died_of['killed_by'], top_weapons_erg)].copy()
top_weapon_kills['bin'] = pd.cut(top_weapon_kills['erg_dist'], np.arange(0, 800, 10), include_lowest=True, labels=False)
top_weapon_kills_wide = top_weapon_kills.groupby(['killed_by', 'bin']).size().unstack(fill_value=0).transpose()

top_weapon_kills_wide = top_weapon_kills_wide.div(top_weapon_kills_wide.sum(axis=1), axis=0)



def stacked(df):
df_top = df.cumsum(axis=1)
df_bottom = df_top.shift(axis=1).fillna(0)[::-1]
df_stack = pd.concat([df_bottom, df_top], ignore_index=True)
return df_stack

hover = HoverTool(
tooltips=[
("index", "$index"),
("weapon", "@weapon"),
("(x,y)", "($x, $y)")
],
point_policy='follow_mouse'
)

areas = stacked(top_weapon_kills_wide)

colors = brewer['Spectral'][areas.shape[1]]
x2 = np.hstack((top_weapon_kills_wide.index[::-1],
top_weapon_kills_wide.index)) /0.095

TOOLS="pan,wheel_zoom,box_zoom,reset,previewsave"
output_notebook()
p = figure(x_range=(1, 800), y_range=(0, 1), tools=[TOOLS, hover], plot_width=800)
p.grid.minor_grid_line_color = '#eeeeee'

source = ColumnDataSource(data={
'x': [x2] * areas.shape[1],
'y': [areas[c].values for c in areas],
'weapon': list(top_weapon_kills_wide.columns),
'color': colors
})

p.patches('x', 'y', source=source, legend="weapon",
color='color', alpha=0.8, line_color=None)
p.title.text = "Top10武器在各距離下的擊殺百分比（熱情沙漠米拉瑪）"
p.xaxis.axis_label = "擊殺距離（0-800米）"
p.yaxis.axis_label = "擊殺百分比"

show(p)

 

 

自己把自己干倒的方式與人數

 

death1.head()

 

 

 

 kill_by_self = death1.loc[death1['killer_name']==death1['victim_name'], "killed_by"]

kill_by_self.value_counts()[:10].plot.barh()
plt.xlabel("自斃的人數", fontsize=14)
plt.ylabel("自斃的方式", fontsize=14)
plt.title('自己把自己干倒的方式與人數', fontsize=14)
plt.yticks(fontsize=12)

 

 

 

 

# 自己把自己干倒的人數場均百分比
kill_by_self.shape[0]/death1.shape[0]*100

 

 

 

 完整代碼：

import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
import seaborn as sns
%matplotlib inline
plt.rcParams['font.sans-serif'] = ['SimHei'] # 指定默認字體
plt.rcParams['axes.unicode_minus'] = False # 解決保存圖像是負號'-'顯示為方塊的問題

# 使用pandas讀取數據
agg1 = pd.read_csv('/Users/apple/Desktop/pubg/aggregate/agg_match_stats_1.csv')

# 探索數據結構
agg1.head()

# 總共有13844275行玩家數據，15列
agg1.shape

agg1.columns

agg1.info()

# 丟棄重復數據
agg1.drop_duplicates(inplace=True)

agg1.loc[1]

# 添加是否成功吃雞列
agg1['won'] = agg1['team_placement'] == 1

# 添加是否搭乘過車輛列
agg1['drove'] = agg1['player_dist_ride'] != 0

agg1.loc[agg1['player_kills'] < 40, ['player_kills', 'won']].groupby('player_kills').won.mean().plot.bar(figsize=(15,6), rot=0)
plt.xlabel('擊殺人數', fontsize=14)
plt.ylabel("吃雞概率", fontsize=14)
plt.title('擊殺人數與吃雞概率的關系', fontsize=14)

agg1.groupby('party_size').player_kills.mean()

g = sns.FacetGrid(agg1.loc[agg1['player_kills']<=10, ['party_size', 'player_kills']], row="party_size", size=4, aspect=2)
g = g.map(sns.countplot, "player_kills")

agg1.loc[agg1['party_size']!=1, ['player_assists', 'won']].groupby('player_assists').won.mean().plot.bar(figsize=(15,6), rot=0)
plt.xlabel('助攻次數', fontsize=14)
plt.ylabel("吃雞概率", fontsize=14)
plt.title('助攻次數與吃雞概率的關系', fontsize=14)

agg1.groupby('drove').won.mean().plot.barh(figsize=(6,3))
plt.xlabel("吃雞概率", fontsize=14)
plt.ylabel("是否搭乘過車輛", fontsize=14)
plt.title('搭乘車輛與吃雞概率的關系', fontsize=14)
plt.yticks([1,0],['是','否'])

dist_ride = agg1.loc[agg1['player_dist_ride']<12000, ['player_dist_ride', 'won']]

labels=["0-1k", "1-2k", "2-3k", "3-4k","4-5k", "5-6k", "6-7k", "7-8k", "8-9k", "9-10k", "10-11k", "11-12k"]
dist_ride['drove_cut'] = pd.cut(dist_ride['player_dist_ride'], 12, labels=labels)

dist_ride.groupby('drove_cut').won.mean().plot.bar(rot=60, figsize=(8,4))
plt.xlabel("搭乘車輛里程", fontsize=14)
plt.ylabel("吃雞概率", fontsize=14)
plt.title('搭乘車輛里程與吃雞概率的關系', fontsize=14)

match_unique = agg1.loc[agg1['party_size'] == 1, 'match_id'].unique()

# 先把玩家被擊殺的數據導入進來並探索數據
death1 = pd.read_csv('/Users/apple/Desktop/pubg/deaths/kill_match_stats_final_1.csv')

death1.head()

death1.info()

death1.shape

death1_solo = death1[death1['match_id'].isin(match_unique)]

death1_solo.info()

# 只統計單人模式，篩選存活不超過180秒的玩家數據
death_180_seconds_erg = death1_solo.loc[(death1_solo['map'] == 'ERANGEL')&(death1_solo['time'] < 180)&(death1_solo['victim_position_x']>0), :].dropna()
death_180_seconds_mrm = death1_solo.loc[(death1_solo['map'] == 'MIRAMAR')&(death1_solo['time'] < 180)&(death1_solo['victim_position_x']>0), :].dropna()

death_180_seconds_erg.shape

death_180_seconds_mrm.shape

# 選擇存活不過180秒的玩家死亡位置
data_erg = death_180_seconds_erg[['victim_position_x', 'victim_position_y']].values
data_mrm = death_180_seconds_mrm[['victim_position_x', 'victim_position_y']].values

# 重新scale玩家位置
data_erg = data_erg*4096/800000
data_mrm = data_mrm*1000/800000

from scipy.ndimage.filters import gaussian_filter
import matplotlib.cm as cm
from matplotlib.colors import Normalize
from scipy.misc.pilutil import imread

from scipy.ndimage.filters import gaussian_filter
import matplotlib.cm as cm
from matplotlib.colors import Normalize

def heatmap(x, y, s, bins=100):
    heatmap, xedges, yedges = np.histogram2d(x, y, bins=bins)
    heatmap = gaussian_filter(heatmap, sigma=s)

    extent = [xedges[0], xedges[-1], yedges[0], yedges[-1]]
    return heatmap.T, extent

bg = imread('/Users/apple/Desktop/pubg/erangel.jpg')
hmap, extent = heatmap(data_erg[:,0], data_erg[:,1], 4.5)
alphas = np.clip(Normalize(0, hmap.max(), clip=True)(hmap)*4.5, 0.0, 1.)
colors = Normalize(0, hmap.max(), clip=True)(hmap)
colors = cm.Reds(colors)
colors[..., -1] = alphas

fig, ax = plt.subplots(figsize=(24,24))
ax.set_xlim(0, 4096); ax.set_ylim(0, 4096)
ax.imshow(bg)
ax.imshow(colors, extent=extent, origin='lower', cmap=cm.Reds, alpha=0.9)
plt.gca().invert_yaxis()

bg = imread('/Users/apple/Desktop/pubg/miramar.jpg')
hmap, extent = heatmap(data_mrm[:,0], data_mrm[:,1], 4)
alphas = np.clip(Normalize(0, hmap.max(), clip=True)(hmap)*4, 0.0, 1.)
colors = Normalize(0, hmap.max(), clip=True)(hmap)
colors = cm.Reds(colors)
colors[..., -1] = alphas

fig, ax = plt.subplots(figsize=(24,24))
ax.set_xlim(0, 1000); ax.set_ylim(0, 1000)
ax.imshow(bg)
ax.imshow(colors, extent=extent, origin='lower', cmap=cm.Reds, alpha=0.9)
#plt.scatter(plot_data_mr[:,0], plot_data_mr[:,1])
plt.gca().invert_yaxis()

death_final_circle_erg = death1_solo.loc[(death1_solo['map'] == 'ERANGEL')&(death1_solo['victim_placement'] == 2)&(death1_solo['victim_position_x']>0)&(death1_solo['killer_position_x']>0), :].dropna()
death_final_circle_mrm = death1_solo.loc[(death1_solo['map'] == 'MIRAMAR')&(death1_solo['victim_placement'] == 2)&(death1_solo['victim_position_x']>0)&(death1_solo['killer_position_x']>0), :].dropna()

print(death_final_circle_erg.shape)
print(death_final_circle_mrm.shape)

final_circle_erg = np.vstack([death_final_circle_erg[['victim_position_x', 'victim_position_y']].values, 
                                    death_final_circle_erg[['killer_position_x', 'killer_position_y']].values])*4096/800000
final_circle_mrm = np.vstack([death_final_circle_mrm[['victim_position_x', 'victim_position_y']].values,
                                    death_final_circle_mrm[['killer_position_x', 'killer_position_y']].values])*1000/800000

bg = imread('/Users/apple/Desktop/pubg/erangel.jpg')
hmap, extent = heatmap(final_circle_erg[:,0], final_circle_erg[:,1], 1.5)
alphas = np.clip(Normalize(0, hmap.max(), clip=True)(hmap)*1.5, 0.0, 1.)
colors = Normalize(0, hmap.max(), clip=True)(hmap)
colors = cm.Reds(colors)
colors[..., -1] = alphas

fig, ax = plt.subplots(figsize=(24,24))
ax.set_xlim(0, 4096); ax.set_ylim(0, 4096)
ax.imshow(bg)
ax.imshow(colors, extent=extent, origin='lower', cmap=cm.Reds, alpha=0.9)
#plt.scatter(plot_data_er[:,0], plot_data_er[:,1])
plt.gca().invert_yaxis()

bg = imread('/Users/apple/Desktop/pubg/miramar.jpg')
hmap, extent = heatmap(final_circle_mrm[:,0], final_circle_mrm[:,1], 1.5)
alphas = np.clip(Normalize(0, hmap.max(), clip=True)(hmap)*1.5, 0.0, 1.)
colors = Normalize(0, hmap.max(), clip=True)(hmap)
colors = cm.Reds(colors)
colors[..., -1] = alphas

fig, ax = plt.subplots(figsize=(24,24))
ax.set_xlim(0, 1000); ax.set_ylim(0, 1000)
ax.imshow(bg)
ax.imshow(colors, extent=extent, origin='lower', cmap=cm.Reds, alpha=0.9)
#plt.scatter(plot_data_mr[:,0], plot_data_mr[:,1])
plt.gca().invert_yaxis()

erg_died_of = death1.loc[(death1['map']=='ERANGEL')&(death1['killer_position_x']>0)&(death1['victim_position_x']>0)&(death1['killed_by']!='Down and Out'),:]
mrm_died_of = death1.loc[(death1['map']=='MIRAMAR')&(death1['killer_position_x']>0)&(death1['victim_position_x']>0)&(death1['killed_by']!='Down and Out'),:]

print(erg_died_of.shape)
print(mrm_died_of.shape)

erg_died_of['killed_by'].value_counts()[:10].plot.barh(figsize=(10,5))
plt.xlabel("被擊殺人數", fontsize=14)
plt.ylabel("擊殺的武器", fontsize=14)
plt.title('武器跟擊殺人數的統計(絕地海島艾倫格)', fontsize=14)
plt.yticks(fontsize=12)

mrm_died_of['killed_by'].value_counts()[:10].plot.barh(figsize=(10,5))
plt.xlabel("被擊殺人數", fontsize=14)
plt.ylabel("擊殺的武器", fontsize=14)
plt.title('武器跟擊殺人數的統計(熱情沙漠米拉瑪)', fontsize=14)
plt.yticks(fontsize=12)

# 把位置信息轉換成距離，以“米”為單位
erg_distance = np.sqrt(((erg_died_of['killer_position_x']-erg_died_of['victim_position_x'])/100)**2 + ((erg_died_of['killer_position_y']-erg_died_of['victim_position_y'])/100)**2)

mrm_distance = np.sqrt(((mrm_died_of['killer_position_x']-mrm_died_of['victim_position_x'])/100)**2 + ((mrm_died_of['killer_position_y']-mrm_died_of['victim_position_y'])/100)**2)

sns.distplot(erg_distance.loc[erg_distance<400])

erg_died_of.loc[(erg_distance > 800)&(erg_distance < 1500), 'killed_by'].value_counts()[:10].plot.bar(rot=30)
plt.xlabel("狙擊的武器", fontsize=14)
plt.ylabel("被狙擊的人數", fontsize=14)
plt.title('狙擊武器跟擊殺人數的統計(絕地海島艾倫格)', fontsize=14)
plt.yticks(fontsize=12)

mrm_died_of.loc[(mrm_distance > 800)&(mrm_distance < 1000), 'killed_by'].value_counts()[:10].plot.bar(rot=30)
plt.xlabel("狙擊的武器", fontsize=14)
plt.ylabel("被狙擊的人數", fontsize=14)
plt.title('狙擊武器跟擊殺人數的統計(熱情沙漠米拉瑪)', fontsize=14)
plt.yticks(fontsize=12)

erg_died_of.loc[erg_distance<10, 'killed_by'].value_counts()[:10].plot.bar(rot=30)
plt.xlabel("近戰武器", fontsize=14)
plt.ylabel("被擊殺的人數", fontsize=14)
plt.title('近戰武器跟擊殺人數的統計(絕地海島艾倫格)', fontsize=14)
plt.yticks(fontsize=12)

mrm_died_of.loc[mrm_distance<10, 'killed_by'].value_counts()[:10].plot.bar(rot=30)
plt.xlabel("近戰武器", fontsize=14)
plt.ylabel("被擊殺的人數", fontsize=14)
plt.title('近戰武器武器跟擊殺人數的統計(熱情沙漠米拉瑪)', fontsize=14)
plt.yticks(fontsize=12)

erg_died_of['erg_dist'] = erg_distance
erg_died_of = erg_died_of.loc[erg_died_of['erg_dist']<800, :]
top_weapons_erg = list(erg_died_of['killed_by'].value_counts()[:10].index)
top_weapon_kills = erg_died_of[np.in1d(erg_died_of['killed_by'], top_weapons_erg)].copy()
top_weapon_kills['bin'] = pd.cut(top_weapon_kills['erg_dist'], np.arange(0, 800, 10), include_lowest=True, labels=False)
top_weapon_kills_wide = top_weapon_kills.groupby(['killed_by', 'bin']).size().unstack(fill_value=0).transpose()

top_weapon_kills_wide = top_weapon_kills_wide.div(top_weapon_kills_wide.sum(axis=1), axis=0)

from bokeh.models.tools import HoverTool
from bokeh.palettes import brewer
from bokeh.plotting import figure, show, output_notebook
from bokeh.models.sources import ColumnDataSource

def  stacked(df):
    df_top = df.cumsum(axis=1)
    df_bottom = df_top.shift(axis=1).fillna(0)[::-1]
    df_stack = pd.concat([df_bottom, df_top], ignore_index=True)
    return df_stack

hover = HoverTool(
    tooltips=[
            ("index", "$index"),
            ("weapon", "@weapon"),
            ("(x,y)", "($x, $y)")
        ],
    point_policy='follow_mouse'
    )

areas = stacked(top_weapon_kills_wide)

colors = brewer['Spectral'][areas.shape[1]]
x2 = np.hstack((top_weapon_kills_wide.index[::-1],
                top_weapon_kills_wide.index)) /0.095

TOOLS="pan,wheel_zoom,box_zoom,reset,previewsave"
output_notebook()
p = figure(x_range=(1, 800), y_range=(0, 1), tools=[TOOLS, hover], plot_width=800)
p.grid.minor_grid_line_color = '#eeeeee'

source = ColumnDataSource(data={
    'x': [x2] * areas.shape[1],
    'y': [areas[c].values for c in areas],
    'weapon': list(top_weapon_kills_wide.columns),
    'color': colors
})

p.patches('x', 'y', source=source, legend="weapon",
          color='color', alpha=0.8, line_color=None)
p.title.text = "Top10武器在各距離下的擊殺百分比（絕地海島艾倫格）"
p.xaxis.axis_label = "擊殺距離（0-800米）"
p.yaxis.axis_label = "百分比"

show(p)

mrm_died_of['erg_dist'] = mrm_distance
mrm_died_of = mrm_died_of.loc[mrm_died_of['erg_dist']<800, :]
top_weapons_erg = list(mrm_died_of['killed_by'].value_counts()[:10].index)
top_weapon_kills = mrm_died_of[np.in1d(mrm_died_of['killed_by'], top_weapons_erg)].copy()
top_weapon_kills['bin'] = pd.cut(top_weapon_kills['erg_dist'], np.arange(0, 800, 10), include_lowest=True, labels=False)
top_weapon_kills_wide = top_weapon_kills.groupby(['killed_by', 'bin']).size().unstack(fill_value=0).transpose()

top_weapon_kills_wide = top_weapon_kills_wide.div(top_weapon_kills_wide.sum(axis=1), axis=0)
def  stacked(df):
    df_top = df.cumsum(axis=1)
    df_bottom = df_top.shift(axis=1).fillna(0)[::-1]
    df_stack = pd.concat([df_bottom, df_top], ignore_index=True)
    return df_stack

hover = HoverTool(
    tooltips=[
            ("index", "$index"),
            ("weapon", "@weapon"),
            ("(x,y)", "($x, $y)")
        ],
    point_policy='follow_mouse'
    )

areas = stacked(top_weapon_kills_wide)

colors = brewer['Spectral'][areas.shape[1]]
x2 = np.hstack((top_weapon_kills_wide.index[::-1],
                top_weapon_kills_wide.index)) /0.095

TOOLS="pan,wheel_zoom,box_zoom,reset,previewsave"
output_notebook()
p = figure(x_range=(1, 800), y_range=(0, 1), tools=[TOOLS, hover], plot_width=800)
p.grid.minor_grid_line_color = '#eeeeee'

source = ColumnDataSource(data={
    'x': [x2] * areas.shape[1],
    'y': [areas[c].values for c in areas],
    'weapon': list(top_weapon_kills_wide.columns),
    'color': colors
})

p.patches('x', 'y', source=source, legend="weapon",
          color='color', alpha=0.8, line_color=None)
p.title.text = "Top10武器在各距離下的擊殺百分比（熱情沙漠米拉瑪）"
p.xaxis.axis_label = "擊殺距離（0-800米）"
p.yaxis.axis_label = "擊殺百分比"

show(p)

death1.head()

kill_by_self = death1.loc[death1['killer_name']==death1['victim_name'], "killed_by"]

kill_by_self.value_counts()[:10].plot.barh()
plt.xlabel("自斃的人數", fontsize=14)
plt.ylabel("自斃的方式", fontsize=14)
plt.title('自己把自己干倒的方式與人數', fontsize=14)
plt.yticks(fontsize=12)

# 自己把自己干倒的人數場均百分比
kill_by_self.shape[0]/death1.shape[0]*100

五，總結

　　使用matplotlib庫做出游戲中的熱感應圖，是本次程序設計最大的收獲，已經達到預期目標對絕地求生比賽數據集分析和可視化。