[OPGG] 데이터 시각화 및 apply

1. 시각화

import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
import seaborn as sns
import matplotlib as mpl

import requests

import warnings
warnings.filterwarnings("ignore")

%matplotlib inline
%config InlineBackend.figure_format = 'retina'

mpl.rc('font', family='NanumGothic') # 폰트 설정
mpl.rc('axes', unicode_minus=False) # 유니코드에서 음수 부호 설정

# 차트 스타일 설정
sns.set(font="NanumGothic", rc={"axes.unicode_minus":False}, style='darkgrid')
plt.rc("figure", figsize=(10,8))

plt.plot([1, 2, 3, 5 ,8])
plt.ylabel("numbers")
plt.show()

# subplot
alphabet = ["a", "b", "c"]
values = [1, 50, 100]

plt.figure(figsize=(12,4))

plt.subplot(131)
plt.bar(alphabet, values)

plt.subplot(1,3,2)
plt.scatter(alphabet, values)

plt.subplot(1,3,3)
plt.plot(alphabet, values)

plt.suptitle("Categorical Plotting")
plt.show()

data = pd.read_csv("Day09_01_BIPA_data.csv", index_col=0)

# 일부 챔피언에 대한 평균값 정보 (10개가 아닌 9개의 챔피언 id 0은 없음)
data1_10 = data[data['championId'].isin(range(0,10))]
group1_10 = data1_10.groupby("championId", as_index=False).mean()

# 챔피언별 총 딜량
fig, ax = plt.subplots(figsize=(7,4))

ax.barh(group1_10['championId'], group1_10['totalDamageDealtToChampions'])
fig.show()

# 각 버전 (챔피언, 룬, 아이템 등..)
champ_ver = requests.get('https://ddragon.leagueoflegends.com/realms/na.json').json()['n']['champion']
championJsonURL = 'http://ddragon.leagueoflegends.com/cdn/' + champ_ver + '/data/en_US/champion.json'

# 챔피언 정보 url
request = requests.get(championJsonURL)
champion_data = request.json()
champion_data.keys()

dict_keys(['type', 'format', 'version', 'data'])

# 챔피언 id, name 데이터 프레임
champion_dict = {}

for c_name in champion_data['data'].keys() :
    champion_dict[int(champion_data['data'][c_name]['key'])]=c_name
    
champion = pd.DataFrame.from_dict(champion_dict, orient = 'index', columns = ['champion'])

# merge, 챔피언 이름 추가
group1_10 = pd.merge(group1_10, champion, left_on="championId", right_index=True)

# 챔피언 이름으로 y축 변경
fig, ax = plt.subplots(figsize=(7,4))

ax.barh(group1_10['champion'], group1_10['totalDamageDealtToChampions'])
fig.show()

# pandas에서 plot
group1_10['totalDamageDealtToChampions'].plot(kind='barh')
plt.show()

# index 변경
group1_10.index = group1_10.champion
group1_10['totalDamageDealtToChampions'].plot(kind='barh')
plt.show()

# color 스타일 변경 (영구)
sns.set_palette('bright')

fig, ax = plt.subplots(figsize=(7,4))

ax.barh(group1_10['champion'], group1_10['totalDamageDealtToChampions'])
fig.show()

# color 변경
sns.set_palette('bright')

fig, ax = plt.subplots(figsize=(7,4))

ax.barh(group1_10['champion'], group1_10['totalDamageDealtToChampions'], color="red")
fig.show()

# seaborn
fig, ax = plt.subplots(figsize=(7,4))

sns.barplot(data=group1_10, x='totalDamageDealtToChampions', y='champion')
plt.show()

# xlabel, ylabel, title
fig, ax = plt.subplots(figsize=(7,4))

sns.barplot(data=group1_10, x='totalDamageDealtToChampions', y='champion')

ax.set_xlabel("Avg_damage")
ax.set_ylabel("Champion Name")
ax.set_title("Avg Champion Damage")

plt.show()

# font style
font_label = {
    'color': 'blue',
    'weight': 'bold'
}

font_title = {
    'family': 'serif',
    'size': 20,
    'backgroundcolor': 'y',
    'weight': 'bold',
    'verticalalignment': 'baseline',
    'horizontalalignment': 'center'
}

fig, ax = plt.subplots(figsize=(7,4))

sns.barplot(data=group1_10, x='totalDamageDealtToChampions', y='champion')

# pad 옵션은 축과의 거리 옵션
ax.set_xlabel("Avg_damage", fontdict=font_label, labelpad=20)
ax.set_ylabel("Champion Name", color='blue', labelpad=100)
ax.set_title("Avg Champion Damage", fontdict=font_title, pad=12)

plt.show()

# font style
font_label = {
    'color': 'blue',
    'weight': 'bold'
}

font_title = {
    'family': 'serif',
    'size': 20,
    'backgroundcolor': 'y',
    'weight': 'bold',
    'verticalalignment': 'baseline',
    'horizontalalignment': 'center'
}

fig, ax = plt.subplots(figsize=(7,4))

sns.barplot(data=group1_10, x='totalDamageDealtToChampions', y='champion')

# pad 옵션은 축과의 거리 옵션
ax.set_xlabel("Avg_damage", fontdict=font_label, labelpad=20)
ax.set_ylabel("Champion Name", color='blue', labelpad=100)
ax.set_title("Avg Champion Damage", fontdict=font_title, pad=12)

# x축 범위 설정
ax.set_xlim(10000, 20000)

plt.show()

# font style
font_label = {
    'color': 'blue',
    'weight': 'bold'
}

font_title = {
    'family': 'serif',
    'size': 20,
    'backgroundcolor': 'y',
    'weight': 'bold',
    'verticalalignment': 'baseline',
    'horizontalalignment': 'center'
}

fig, ax = plt.subplots(figsize=(7,4))

sns.barplot(data=group1_10, x='totalDamageDealtToChampions', y='champion')

# pad 옵션은 축과의 거리 옵션
ax.set_xlabel("Avg_damage", fontdict=font_label, labelpad=20)
ax.set_ylabel("Champion Name", color='blue', labelpad=100)
ax.set_title("Avg Champion Damage", fontdict=font_title, pad=12)

# x축 범위 설정
ax.set_xlim(10000, 20000)
ax.axvline(damage_mean, ls='--', lw=1, color='green')
# 평균값 선 추가해주기
damage_mean = group1_10['totalDamageDealtToChampions'].mean()
ax.axvline(damage_mean, ls='--', lw=1, color='green')

# 화살표 추가
arrowprops = {
    'arrowstyle': '->'
}
ax.annotate("average", (damage_mean, 2.5), xytext=(17000,2.5), color='green',
            fontfamily='serif', fontstyle='italic', fontsize=15, arrowprops = arrowprops)

plt.show()

# boxplot
sns.boxplot(y = data['totalDamageDealtToChampions'], x = data['position'])
plt.show()

# boxplot
sns.violinplot(y = data['totalDamageDealtToChampions'], x = data['position'])
plt.show()

# distplot
sns.distplot(data['totalDamageDealtToChampions'])
plt.show()

# distplot
group_data = data[data['gameLength']>200].groupby(['position','championId'], as_index=False).mean()
# sns.distplot(group_data, x='totalDamageDealtToChampions', hue='position')
sns.distplot(group_data['totalDamageDealtToChampions'])
plt.show()

# displot
# distplot은 figure size가 안먹혀서 displot으로 heigt, aspect로 설정가능
sns.displot(group_data['totalDamageDealtToChampions'], height=4, aspect=2)
plt.show()

# displot
sns.displot(group_data[group_data['position'].isin(['A','S'])], 
            x = 'totalDamageDealtToChampions', hue='position', multiple='dodge', height=4, aspect=2)
plt.show()

# 챔피언별 level당 stat 정보 등
champ_stats = pd.read_csv('Day09_01_champ_stats.csv')

champ_stats

	championId	championName	versionIndex	version	hp	hpperlevel	mp	mpperlevel	movespeed	armor	...	attackspeed	hp_18	mp_18	armor_18	spellblock_18	hpregen_18	mpregen_18	crit_18	attackdamage_18	attackspeed_18
0	1	Annie	11.17	11.17.394.4489	524.0	88	418.0	25.0	335	19	...	0.579	2020.0	843.0	87.00	38.50	14.85	21.60	0	94.71	0.7129
1	2	Olaf	11.17	11.17.394.4489	575.0	100	316.0	42.0	350	35	...	0.694	2275.0	1030.0	86.00	53.25	23.80	17.70	0	127.50	1.0125
2	3	Galio	11.17	11.17.394.4489	562.0	112	500.0	40.0	335	24	...	0.625	2466.0	1180.0	83.50	53.25	21.60	21.40	0	118.50	0.7844
3	4	TwistedFate	11.17	11.17.394.4489	534.0	94	333.0	39.0	330	21	...	0.651	2132.0	996.0	74.55	38.50	15.70	21.60	0	108.10	1.0074
4	5	XinZhao	11.17	11.17.394.4489	570.0	92	274.0	55.0	345	35	...	0.645	2134.0	1209.0	94.50	53.25	19.90	14.91	0	114.00	1.0288
...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...
151	555	Pyke	11.17	11.17.394.4489	600.0	90	415.0	50.0	330	45	...	0.667	2130.0	1265.0	104.50	53.25	15.50	25.00	0	96.00	0.9505
152	777	Yone	11.17	11.17.394.4489	550.0	85	500.0	0.0	345	28	...	0.625	1995.0	500.0	85.80	53.25	20.25	0.00	0	94.00	0.9969
153	875	Sett	11.17	11.17.394.4489	600.0	93	0.0	0.0	340	33	...	0.625	2181.0	0.0	101.00	53.25	15.50	0.00	0	128.00	0.8109
154	876	Lillia	11.17	11.17.394.4489	580.0	90	410.0	50.0	330	22	...	0.625	2110.0	1260.0	90.00	44.75	20.25	27.65	0	113.70	0.9119
155	887	Gwen	11.17	11.17.394.4489	550.0	90	330.0	40.0	340	39	...	0.690	2080.0	1010.0	107.00	53.25	16.35	19.40	0	114.00	0.9539

156 rows × 33 columns

# scatterplot
sns.scatterplot(data=champ_stats, x='hp', y='hp_18')
plt.show()

champ_stats.columns

Index(['championId', 'championName', 'versionIndex', 'version', 'hp',
       'hpperlevel', 'mp', 'mpperlevel', 'movespeed', 'armor', 'armorperlevel',
       'spellblock', 'spellblockperlevel', 'attackrange', 'hpregen',
       'hpregenperlevel', 'mpregen', 'mpregenperlevel', 'crit', 'critperlevel',
       'attackdamage', 'attackdamageperlevel', 'attackspeedperlevel',
       'attackspeed', 'hp_18', 'mp_18', 'armor_18', 'spellblock_18',
       'hpregen_18', 'mpregen_18', 'crit_18', 'attackdamage_18',
       'attackspeed_18'],
      dtype='object')

# text 삽입
sns.scatterplot(data=champ_stats, x='hp', y='hp_18')

plt.text(x = champ_stats[champ_stats['championName'] == 'Kled']['hp'] + 3,
         y = champ_stats[champ_stats['championName'] == 'Kled']['hp_18'] + 3,
         s = 'Kled')

plt.show()

# 모든 챔피언 이름 넣기
sns.scatterplot(data=champ_stats, x='hp', y='hp_18')

for name in champ_stats['championName'].values:
    plt.text(x = champ_stats[champ_stats['championName'] == name]['hp'] + 3,
             y = champ_stats[champ_stats['championName'] == name]['hp_18'] + 3,
             s = name)

plt.show()

# 공격력과 공격속도
sns.scatterplot(data=champ_stats, x='attackdamage', y='attackspeed')

for name in champ_stats['championName'].values:
    plt.text(x = champ_stats[champ_stats['championName'] == name]['attackdamage'] + 0.5,
             y = champ_stats[champ_stats['championName'] == name]['attackspeed'] + 0.01,
             s = name)

plt.show()

2. 함수 정의 및 apply

# sample data
data = pd.DataFrame([[1,2,3,4], [5,6,7,8], [9,8,7,6]],
                    index = ['a', 'b', 'c'], columns = ["A", "B", "C", "D"])

data

	A	B	C	D
a	1	2	3	4
b	5	6	7	8
c	9	8	7	6

# 열 단위
data.apply(np.mean, axis=0)

A    5.000000
B    5.333333
C    5.666667
D    6.000000
dtype: float64

# 행 단위
data.apply(np.mean, axis=1)

a    2.5
b    6.5
c    7.5
dtype: float64

# exponential
data.apply(np.exp)

	A	B	C	D
a	2.718282	7.389056	20.085537	54.598150
b	148.413159	403.428793	1096.633158	2980.957987
c	8103.083928	2980.957987	1096.633158	403.428793

data.apply(lambda x: x["A"] + x["B"], axis=1)

a     3
b    11
c    17
dtype: int64

data.apply(lambda x: x["a"] + x["c"], axis=0)

A    10
B    10
C    10
D    10
dtype: int64

# 함수 이용
def multiply_2(x):
    return x*2

data.apply(lambda x: multiply_2(x))

	A	B	C	D
a	2	4	6	8
b	10	12	14	16
c	18	16	14	12

# 함수 이용2
def hello(x):
    return 'hello my name is ' + x

hello("Me")

'hello my name is Me'

data = pd.read_csv("Day09_01_BIPA_data.csv", index_col=0)

data

	gameId	createDate	tierRank	position	teamId	summonerId	championId	result	level	championsKilled	...	totalDamageTaken	neutralMinionsKilledEnemyJungle	neutralMinionsKilledTeamJungle	visionScore	damageSelfMitigated	damageDealtToObjectives	damageDealtToTurrets	lane	gameLength	version
0	5383880006	2021-08-11 06:59:31	P318	S	200	3210877	99	WIN	10	4	...	3602	0	0	11	2830	2181	2181	NONE	972	11.16.390.1945
1	5383880006	2021-08-11 06:59:31	P466	S	100	7550211	35	LOSE	8	1	...	5077	0	2	24	2013	0	0	NONE	972	11.16.390.1945
2	5383880006	2021-08-11 06:59:31	P329	M	200	9306696	266	WIN	11	5	...	11029	0	0	15	6158	585	585	NONE	972	11.16.390.1945
3	5383880006	2021-08-11 06:59:31	P461	T	200	20281103	92	WIN	11	8	...	9944	0	0	10	7615	2318	2318	NONE	972	11.16.390.1945
4	5383880006	2021-08-11 06:59:31	P30	A	200	26082075	22	WIN	10	3	...	6387	0	0	13	3218	2370	2370	NONE	972	11.16.390.1945
...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...
999995	5384659633	2021-08-11 17:20:44	S146	S	100	21560397	53	WIN	15	1	...	32961	0	0	63	34725	1978	126	BOTTOM	2379	11.16.390.1945
999996	5384659633	2021-08-11 17:20:44	G442	M	100	46320531	166	WIN	18	10	...	24674	1	16	16	21371	8406	7042	MIDDLE	2379	11.16.390.1945
999997	5384659633	2021-08-11 17:20:44	G418	J	200	64820251	64	LOSE	17	5	...	45911	6	68	22	62426	20292	454	JUNGLE	2379	11.16.390.1945
999998	5384659633	2021-08-11 17:20:44	S178	A	100	72310857	81	WIN	17	6	...	28187	8	9	17	19664	11090	1062	BOTTOM	2379	11.16.390.1945
999999	5384659633	2021-08-11 17:20:44	G456	T	100	72520226	875	WIN	18	3	...	54999	0	9	30	64441	13682	9331	TOP	2379	11.16.390.1945

1000000 rows × 37 columns

# 첫 글자 가져오기
data["tierRank"].str[:1]

0         P
1         P
2         P
3         P
4         P
         ..
999995    S
999996    G
999997    G
999998    S
999999    G
Name: tierRank, Length: 1000000, dtype: object

# 첫 글자 가져오기 apply
# data["tierRank"].apply(lambda x: x[0])

• 실행시 결측 값이 존재해서 float형태로 입력되어 있기에 인덱싱이 안됨(문자열 x)

sns.heatmap(data.isnull(), cbar=False)
plt.show()

# missingno 패키지 사용
import missingno as msno

msno.bar(data)

<AxesSubplot:>

• tierRank에 결측값 존재하는 것 확인 가능

# 결측값 제거
data1 = data[data["tierRank"].notnull()].copy()

# 첫 글자 가져오기 apply
data1["tier"] = data1["tierRank"].apply(lambda x: x[0])

• 이제 float형태가 아니므로 잘 적용된다.

# 특정 문자가 있는지
data1[data1["tierRank"].str.contains("C")]

	gameId	createDate	tierRank	position	teamId	summonerId	championId	result	level	championsKilled	...	neutralMinionsKilledEnemyJungle	neutralMinionsKilledTeamJungle	visionScore	damageSelfMitigated	damageDealtToObjectives	damageDealtToTurrets	lane	gameLength	version	tier
36518	5383988789	2021-08-11 11:21:27	C1907	T	200	86094085	39	WIN	16	6	...	1	1	26	21058	13819	10387	TOP	1592	11.16.390.1945	C
205196	5384285152	2021-08-11 12:34:20	C1819	J	100	87214832	234	LOSE	12	2	...	1	101	34	16263	4647	0	JUNGLE	1531	11.16.390.1945	C
206972	5384286004	2021-08-11 12:43:57	C1868	J	100	41780873	68	WIN	15	12	...	20	100	37	17796	32194	1021	JUNGLE	1612	11.16.390.1945	C
210966	5384288138	2021-08-11 13:09:53	C1906	M	100	86094085	13	WIN	18	8	...	12	15	31	11095	26446	8164	TOP	1859	11.16.390.1945	C
210969	5384288138	2021-08-11 13:09:53	C1843	T	200	87299019	150	LOSE	17	5	...	0	3	26	27323	5388	0	TOP	1859	11.16.390.1945	C
...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...
944443	5384634868	2021-08-11 16:27:29	C1836	J	200	84679870	234	LOSE	10	2	...	17	58	20	8843	12620	0	NONE	1148	11.16.390.1945	C
950550	5384637692	2021-08-11 16:57:07	C1918	J	100	6300872	131	WIN	16	5	...	20	100	20	19966	43562	2366	JUNGLE	1774	11.16.390.1945	C
950551	5384637692	2021-08-11 16:57:07	C1857	S	200	13583512	48	LOSE	13	2	...	0	0	70	20091	2733	1103	BOTTOM	1774	11.16.390.1945	C
950554	5384637692	2021-08-11 16:57:07	C1864	M	200	79280308	7	LOSE	17	19	...	0	16	32	13141	10980	8699	MIDDLE	1774	11.16.390.1945	C
950557	5384637692	2021-08-11 16:57:07	C1806	T	200	87299019	223	LOSE	14	3	...	0	0	16	55072	4999	2161	TOP	1774	11.16.390.1945	C

211 rows × 38 columns

data1["tier"].value_counts()

S    350067
G    332395
B    127999
P    125914
D     26826
I     10052
M      2965
R       712
C       211
Name: tier, dtype: int64

# 티어별 카테고리 생성
data1["tier_category"] = data1['tier'].apply(lambda x: 1 if x=="I"
                                             else 2 if x=="B"
                                             else 3 if x=="S"
                                             else 4 if x=="G"
                                             else 5 if x=="P"
                                             else 6 if x=="D"
                                             else 7 if x=="M"
                                             else 8 if x=="R"
                                             else 9 if x=="C"
                                             else 0)

data1[["tier", "tier_category"]]

	tier	tier_category
0	P	5
1	P	5
2	P	5
3	P	5
4	P	5
...	...	...
999995	S	3
999996	G	4
999997	G	4
999998	S	3
999999	G	4

977141 rows × 2 columns

# 티어별 카테고리 생성 2
def tier_category_function(x):
    if x=="I":
        return 1
    if x=="B":
        return 2
    if x=="S":
        return 3
    if x=="G":
        return 4
    if x=="P":
        return 5
    if x=="D":
        return 6
    if x=="M":
        return 7
    if x=="R":
        return 8
    if x=="C":
        return 9

data1["tier_category2"] = data1['tier'].apply(lambda x: tier_category_function(x))

len(data1["tier_category"] == data1["tier_category2"])

977141

# 내가 만든 함수 불러오기
from Day09_01_module import champion_load
champion_load()

	champion
266	Aatrox
103	Ahri
84	Akali
166	Akshan
12	Alistar
...	...
238	Zed
115	Ziggs
26	Zilean
142	Zoe
143	Zyra

156 rows × 1 columns

• 미리 만들어둔 Day09_01_module 파일에 champion_load 함수를 불러와 사용하였다.

3. 와드와 승률의 상관관계

# 승리여부 1,0으로 변환
data_me = data[["wardPlaced", "result"]]
data_me["result"] = data_me["result"].apply(lambda x: 1 if x=="WIN" else 0)

# 승률 계산
def count_sum(x):
    win_rate = x['result'].sum() / x['result'].count()
    
    return pd.Series(win_rate, index=["win_rate"])

temp = data_me.groupby("wardPlaced", as_index=False).apply(count_sum)

# 와드수 40개 이하만
temp_ward40 = temp[temp["wardPlaced"] <= 40]

# bar chart
fig, axs = plt.subplots(1,1, figsize=(15,6))

sns.barplot(data = temp_ward40, x = "wardPlaced", y = "win_rate", ci = None, ax= axs)
axs.axhline(0.5, ls='--', lw=1, color='black')
axs.set_ylim(0,0.8)

plt.show()