8-1. 원자재가격과 주가의 관계 분석_1 [Code]
![8-1. 원자재가격과 주가의 관계 분석_1 [Code]](https://github.com/HwangToeMat/HwangToeMat.github.io/blob/master/assets/img/thumbnail/dwp-9.jpg?raw=true)
각종 원자재 가격과 주가의 상관관계를 알아보기위해 5년간 KOSPI 주가의 등락과 원자재가격의 등락을 비교하고 해당기업과 원자재 간의 관계를 해석해 보았다.
원자재가격과 주가의 관계 분석
import pandas as pd
from pandas import Series,DataFrame
import numpy as np
from dateutil.parser import parse
import dateutil
# For Visualization
import matplotlib.pyplot as plt
from matplotlib import pyplot
import seaborn as sns
sns.set_style('whitegrid')
%matplotlib inline
# For time stamps
from datetime import datetime
import warnings
warnings.filterwarnings("ignore")
import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
%matplotlib inline
from datetime import datetime
path = "c:/Windows/Fonts/malgun.ttf"
import platform
from matplotlib import font_manager, rc
if platform.system() == 'Darwin':
rc('font', family='AppleGothic')
elif platform.system() == 'Windows':
font_name = font_manager.FontProperties(fname=path).get_name()
rc('font', family=font_name)
else:
print('Unknown system... sorry~~~~')
plt.rcParams['axes.unicode_minus'] = False
데이터 전처리
KOSPI만 선택
item_Charge = pd.read_csv("st_data_itemChargeFull.tsv", sep = '\t')
item_Charge['date'] = item_Charge['date'].apply(str)
item_Charge['date'] = item_Charge['date'].apply(dateutil.parser.parse)
KOSPI = item_Charge.loc[item_Charge[list(item_Charge.columns)[2]]=="KOSPI",:]
KOSPI.head()
| itemcode | itemname | market | date | close_val | change_val | change_rate | acc_quant | debt_total | face_val | ... | operating_profit | operating_profit_increasing_rate | prev_quant | property_total | reserve_ratio | roa | roe | sales | sales_increasing_rate | trade_stop_yn | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 0 | 000020 | 동화약품 | KOSPI | 2015-01-07 | 5540.0 | 70.0 | 1.26 | 42517.0 | 732.0 | 1000 | ... | 112.0 | NaN | 62744.0 | 3706.0 | 967.43 | NaN | NaN | 3066.0 | NaN | N |
| 1 | 000020 | 동화약품 | KOSPI | 2015-01-13 | 5600.0 | 40.0 | 0.71 | 39527.0 | 732.0 | 1000 | ... | 112.0 | NaN | 62744.0 | 3706.0 | 967.43 | NaN | NaN | 3066.0 | NaN | N |
| 2 | 000020 | 동화약품 | KOSPI | 2015-01-16 | 5640.0 | -110.0 | -1.95 | 57341.0 | 732.0 | 1000 | ... | 112.0 | NaN | 62744.0 | 3706.0 | 967.43 | NaN | NaN | 3066.0 | NaN | N |
| 3 | 000020 | 동화약품 | KOSPI | 2015-01-19 | 5680.0 | 40.0 | 0.70 | 49574.0 | 732.0 | 1000 | ... | 112.0 | NaN | 62744.0 | 3706.0 | 967.43 | NaN | NaN | 3066.0 | NaN | N |
| 4 | 000020 | 동화약품 | KOSPI | 2015-01-22 | 5550.0 | -60.0 | -1.08 | 77386.0 | 732.0 | 1000 | ... | 112.0 | NaN | 62744.0 | 3706.0 | 967.43 | NaN | NaN | 3066.0 | NaN | N |
5 rows × 27 columns
KOSPI_=KOSPI.pivot_table('close_val',index='date',columns='itemname')
KOSPI_clo = KOSPI_.dropna(axis = 1 ).fillna(method='ffill').pct_change().fillna(0)
KOSPI_val = KOSPI_.dropna(axis = 1 ).fillna(method='ffill')
KOSPI_val.head()
| itemname | AJ렌터카 | AK홀딩스 | BGF | BNK금융지주 | BYC | BYC우 | CJ | CJ CGV | CJ대한통운 | CJ씨푸드 | ... | 후성 | 휠라코리아 | 휴니드 | 휴비스 | 휴스틸 | 휴켐스 | 흥국화재 | 흥국화재2우B | 흥국화재우 | 흥아해운 |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| date | |||||||||||||||||||||
| 2015-01-07 | 17750.0 | 86600.0 | 74500.0 | 13500.0 | 252000.0 | 168500.0 | 158500.0 | 59800.0 | 183500.0 | 2595.0 | ... | 3080.0 | 119000.0 | 4510.0 | 11600.0 | 17350.0 | 25550.0 | 3730.0 | 6140.0 | 2460.0 | 2805.0 |
| 2015-01-13 | 17650.0 | 88000.0 | 73700.0 | 14000.0 | 254000.0 | 170000.0 | 161500.0 | 60500.0 | 192000.0 | 2590.0 | ... | 3245.0 | 118000.0 | 4820.0 | 11750.0 | 17600.0 | 25200.0 | 3900.0 | 6200.0 | 2465.0 | 2655.0 |
| 2015-01-16 | 17300.0 | 92300.0 | 75100.0 | 14000.0 | 252500.0 | 169000.0 | 164500.0 | 59300.0 | 182000.0 | 2560.0 | ... | 3215.0 | 115000.0 | 4970.0 | 11800.0 | 17150.0 | 26800.0 | 4230.0 | 6160.0 | 2410.0 | 2710.0 |
| 2015-01-19 | 17250.0 | 88200.0 | 75100.0 | 14250.0 | 252500.0 | 170500.0 | 164500.0 | 59300.0 | 182000.0 | 2540.0 | ... | 3160.0 | 115500.0 | 5050.0 | 11800.0 | 16850.0 | 27050.0 | 4190.0 | 6300.0 | 2450.0 | 2575.0 |
| 2015-01-22 | 17000.0 | 90300.0 | 75000.0 | 13950.0 | 251000.0 | 175000.0 | 168500.0 | 59600.0 | 172000.0 | 2520.0 | ... | 3205.0 | 116000.0 | 5090.0 | 11650.0 | 16800.0 | 26800.0 | 4175.0 | 6180.0 | 2495.0 | 2570.0 |
5 rows × 802 columns
st_data_goodsForeign = pd.read_csv("st_data_goodsForeign.tsv", sep = '\t')
st_data_goodsForeign['date'] = st_data_goodsForeign['date'].apply(str)
st_data_goodsForeign['date'] = st_data_goodsForeign['date'].apply(dateutil.parser.parse)
st_data_goodsForeign.head()
| itemcode | itemname | date | close_val | change_val | open_val | high_val | low_val | |
|---|---|---|---|---|---|---|---|---|
| 0 | LME@AAY | 알루미늄합금 | 2015-01-07 | 1790.0 | -15.0 | 1790.0 | 1790.0 | 1780.0 |
| 1 | LME@AAY | 알루미늄합금 | 2015-01-13 | 1835.0 | NaN | 1835.0 | 1835.0 | 1825.0 |
| 2 | LME@AAY | 알루미늄합금 | 2015-01-16 | 1820.0 | NaN | 1820.0 | 1820.0 | 1810.0 |
| 3 | LME@AAY | 알루미늄합금 | 2015-01-19 | 1820.0 | NaN | 1820.0 | 1820.0 | 1810.0 |
| 4 | LME@AAY | 알루미늄합금 | 2015-01-22 | 1870.0 | 10.0 | 1870.0 | 1870.0 | 1860.0 |
goodsForeign=st_data_goodsForeign.pivot_table('close_val',index='date',columns='itemname')
goodsForeign_clo = goodsForeign.fillna(method='ffill').pct_change().fillna(0)
goodsForeign_val = goodsForeign.fillna(method='ffill')
goodsForeign_val.head()
| itemname | 구리 | 난방유 | 납 | 니켈 | 대두 | 대두박 | 대두유 | 설탕 | 소맥 | 쌀 | 아연 | 알루미늄합금 | 옥수수 | 주석 | 천연가스 |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| date | |||||||||||||||
| 2015-01-07 | 6170.0 | 1.6999 | 1820.0 | 15455.0 | 1065.6 | 368.5 | 32.98 | 14.78 | 579.4 | 11.410 | 2145.5 | 1790.0 | 411.4 | 19680.0 | 2.871 |
| 2015-01-13 | 5915.0 | 1.6330 | 1800.5 | 14780.0 | 1015.6 | 349.7 | 32.39 | 14.87 | 548.0 | 11.240 | 2083.0 | 1835.0 | 400.2 | 19555.0 | 2.943 |
| 2015-01-16 | 5660.0 | 1.6656 | 1766.5 | 14360.0 | 1003.0 | 326.2 | 33.39 | 15.33 | 532.6 | 11.325 | 2056.5 | 1820.0 | 401.0 | 19325.0 | 3.127 |
| 2015-01-19 | 5671.0 | 1.6411 | 1827.5 | 14400.0 | 1003.0 | 326.2 | 33.39 | 15.33 | 532.6 | 11.325 | 2068.5 | 1820.0 | 401.0 | 19280.0 | 2.995 |
| 2015-01-22 | 5742.0 | 1.6379 | 1881.5 | 14720.0 | 988.6 | 330.1 | 31.97 | 15.91 | 533.6 | 11.160 | 2144.0 | 1870.0 | 399.2 | 19580.0 | 2.835 |
TD = pd.concat([KOSPI_val,goodsForeign_val],axis=1)
TD = TD.reset_index().fillna(method='ffill')
for i in range(0,336):
if i%3 != 0:
TD.drop( i,inplace = True)
FTD = TD.set_index('date')
FTD.head()
| itemname | AJ렌터카 | AK홀딩스 | BGF | BNK금융지주 | BYC | BYC우 | CJ | CJ CGV | CJ대한통운 | CJ씨푸드 | ... | 대두박 | 대두유 | 설탕 | 소맥 | 쌀 | 아연 | 알루미늄합금 | 옥수수 | 주석 | 천연가스 |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| date | |||||||||||||||||||||
| 2015-01-07 | 17750.0 | 86600.0 | 74500.0 | 13500.0 | 252000.0 | 168500.0 | 158500.0 | 59800.0 | 183500.0 | 2595.0 | ... | 368.5 | 32.98 | 14.78 | 579.4 | 11.410 | 2145.5 | 1790.0 | 411.4 | 19680.0 | 2.871 |
| 2015-01-19 | 17250.0 | 88200.0 | 75100.0 | 14250.0 | 252500.0 | 170500.0 | 164500.0 | 59300.0 | 182000.0 | 2540.0 | ... | 326.2 | 33.39 | 15.33 | 532.6 | 11.325 | 2068.5 | 1820.0 | 401.0 | 19280.0 | 2.995 |
| 2015-02-03 | 17350.0 | 98000.0 | 85000.0 | 14450.0 | 282000.0 | 171500.0 | 173000.0 | 60200.0 | 200000.0 | 2525.0 | ... | 340.6 | 30.80 | 14.47 | 513.6 | 10.385 | 2146.5 | 1840.0 | 401.0 | 18950.0 | 2.754 |
| 2015-02-12 | 16500.0 | 91500.0 | 88000.0 | 14450.0 | 295000.0 | 170500.0 | 177500.0 | 64700.0 | 197000.0 | 2600.0 | ... | 330.5 | 32.01 | 15.04 | 521.2 | 10.330 | 2129.5 | 1785.0 | 398.2 | 17900.0 | 2.713 |
| 2015-02-27 | 16800.0 | 101000.0 | 89800.0 | 14900.0 | 304500.0 | 186500.0 | 173500.0 | 64300.0 | 181000.0 | 2610.0 | ... | 353.7 | 32.80 | 13.77 | 517.4 | 10.470 | 2060.5 | 1800.0 | 401.2 | 17920.0 | 2.734 |
5 rows × 817 columns
del TD['date']
TD.head()
corr = TD.corr(method = 'pearson')
corr_TD = corr.iloc[:-16,-15:]
corr_TD.head()
| itemname | 구리 | 난방유 | 납 | 니켈 | 대두 | 대두박 | 대두유 | 설탕 | 소맥 | 쌀 | 아연 | 알루미늄합금 | 옥수수 | 주석 | 천연가스 |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| itemname | |||||||||||||||
| AJ렌터카 | 0.253434 | 0.357430 | -0.180675 | 0.491654 | -0.115987 | 0.033026 | -0.198628 | -0.551920 | 0.578182 | 0.134875 | -0.117926 | 0.354760 | 0.391342 | -0.238737 | 0.049452 |
| AK홀딩스 | 0.284143 | 0.319556 | -0.028906 | 0.572011 | 0.177118 | 0.423597 | -0.119239 | -0.449531 | 0.480114 | 0.331932 | -0.006332 | 0.668936 | 0.434461 | -0.125864 | -0.080484 |
| BGF | -0.840062 | -0.796421 | -0.665365 | -0.663681 | 0.135000 | -0.171780 | 0.112896 | 0.434287 | -0.213475 | -0.261622 | -0.742201 | -0.237837 | 0.009903 | -0.649587 | -0.506187 |
| BNK금융지주 | 0.085814 | 0.101259 | -0.238705 | 0.352453 | 0.075016 | 0.130878 | -0.052747 | -0.365024 | 0.446003 | 0.125338 | -0.264136 | 0.536734 | 0.446926 | -0.389719 | -0.135423 |
| BYC | -0.787079 | -0.844172 | -0.445330 | -0.739622 | 0.107104 | -0.327591 | 0.282540 | 0.573646 | -0.448948 | -0.358966 | -0.559299 | -0.355283 | -0.213558 | -0.434541 | -0.579782 |
데이터 시각화
원자재 가격과 양의 상관관계를 갖는 주식
corr_TD.iloc[:,0].sort_values(ascending=False).head()
itemname
POSCO 0.890243
SK하이닉스 0.885247
하나금융지주 0.865733
NH투자증권 0.858859
KB금융 0.854524
Name: 구리, dtype: float64
from sklearn import preprocessing
col = ['POSCO', 'SK하이닉스', '하나금융지주', 'NH투자증권', 'KB금융', '구리']
x = FTD[col].values
min_max_scaler = preprocessing.MinMaxScaler()
x_scaled = min_max_scaler.fit_transform(x.astype(float))
cu = pd.DataFrame(x_scaled, columns = col, index = FTD.index)
cu.head()
| POSCO | SK하이닉스 | 하나금융지주 | NH투자증권 | KB금융 | 구리 | |
|---|---|---|---|---|---|---|
| date | ||||||
| 2015-01-07 | 0.546296 | 0.330261 | 0.309278 | 0.155107 | 0.154161 | 0.644206 |
| 2015-01-19 | 0.513889 | 0.324117 | 0.371134 | 0.274905 | 0.195089 | 0.471332 |
| 2015-02-03 | 0.428241 | 0.319508 | 0.344624 | 0.274905 | 0.234652 | 0.479647 |
| 2015-02-12 | 0.462963 | 0.318740 | 0.243004 | 0.306431 | 0.203274 | 0.479647 |
| 2015-02-27 | 0.490741 | 0.317972 | 0.284242 | 0.438840 | 0.263302 | 0.543911 |
plt.plot(cu['POSCO'],ls = ":", label = 'POSCO')
plt.plot(cu['SK하이닉스'],ls = ":", label = 'SK하이닉스')
plt.plot(cu['하나금융지주'],ls = ":", label = '하나금융지주')
plt.plot(cu['NH투자증권'],ls = ":", label = 'NH투자증권')
plt.plot(cu['KB금융'],ls = ":", label = 'KB금융')
plt.plot(cu['구리'], label = '구리')
plt.legend(loc = 'best')
plt.grid()
plt.show()
plt.savefig('구리.jpg')
<Figure size 432x288 with 0 Axes>
corr_TD.iloc[:,1].sort_values(ascending=False).head()
itemname
금호석유 0.849365
신대양제지 0.792479
SK하이닉스 0.788649
LS산전 0.786274
KG케미칼 0.779705
Name: 난방유, dtype: float64
from sklearn import preprocessing
col = ['금호석유', '신대양제지', 'SK하이닉스', 'LS산전', 'KG케미칼', '난방유']
x = FTD[col].values
min_max_scaler = preprocessing.MinMaxScaler()
x_scaled = min_max_scaler.fit_transform(x.astype(float))
cu = pd.DataFrame(x_scaled, columns = col, index = FTD.index)
cu.head()
| 금호석유 | 신대양제지 | SK하이닉스 | LS산전 | KG케미칼 | 난방유 | |
|---|---|---|---|---|---|---|
| date | ||||||
| 2015-01-07 | 0.508148 | 0.141916 | 0.330261 | 0.486553 | 0.241379 | 0.524168 |
| 2015-01-19 | 0.555556 | 0.112575 | 0.324117 | 0.545232 | 0.295977 | 0.483911 |
| 2015-02-03 | 0.629630 | 0.143713 | 0.319508 | 0.545232 | 0.218391 | 0.624538 |
| 2015-02-12 | 0.540741 | 0.159880 | 0.318740 | 0.559902 | 0.178161 | 0.670546 |
| 2015-02-27 | 0.551111 | 0.171257 | 0.317972 | 0.579462 | 0.221264 | 0.711625 |
plt.plot(cu['금호석유'],ls = ":", label = '금호석유')
plt.plot(cu['신대양제지'],ls = ":", label = '신대양제지')
plt.plot(cu['SK하이닉스'],ls = ":", label = 'SK하이닉스')
plt.plot(cu['LS산전'],ls = ":", label = 'LS산전')
plt.plot(cu['KG케미칼'],ls = ":", label = 'KG케미칼')
plt.plot(cu['난방유'], label = '난방유')
plt.legend(loc = 'best')
plt.grid()
plt.show()
corr_TD.iloc[:,13].sort_values(ascending=False).head()
itemname
카프로 0.838551
LS 0.837574
이구산업 0.816536
POSCO 0.803074
황금에스티 0.779738
Name: 주석, dtype: float64
from sklearn import preprocessing
col = ['카프로', 'LS', '이구산업', 'POSCO', '황금에스티', '주석']
x = FTD[col].values
min_max_scaler = preprocessing.MinMaxScaler()
x_scaled = min_max_scaler.fit_transform(x.astype(float))
cu = pd.DataFrame(x_scaled, columns = col, index = FTD.index)
cu.head()
| 카프로 | LS | 이구산업 | POSCO | 황금에스티 | 주석 | |
|---|---|---|---|---|---|---|
| date | ||||||
| 2015-01-07 | 0.061184 | 0.410669 | 0.062035 | 0.546296 | 0.024415 | 0.741765 |
| 2015-01-19 | 0.218421 | 0.347163 | 0.064516 | 0.513889 | 0.000000 | 0.694706 |
| 2015-02-03 | 0.179605 | 0.359018 | 0.096774 | 0.428241 | 0.018820 | 0.655882 |
| 2015-02-12 | 0.187500 | 0.332769 | 0.114144 | 0.462963 | 0.034588 | 0.532353 |
| 2015-02-27 | 0.228947 | 0.402202 | 0.168734 | 0.490741 | 0.072228 | 0.534706 |
plt.plot(cu['카프로'],ls = ":", label = '카프로')
plt.plot(cu['LS'],ls = ":", label = 'LS')
plt.plot(cu['이구산업'],ls = ":", label = '이구산업')
plt.plot(cu['POSCO'],ls = ":", label = 'POSCO')
plt.plot(cu['황금에스티'],ls = ":", label = '황금에스티')
plt.plot(cu['주석'], label = '주석')
plt.legend(loc = 'best')
plt.grid()
plt.show()
원자재 가격과 음의 상관관계를 갖는 주식
corr_TD.iloc[:,0].sort_values(ascending=True).head()
itemname
미래아이앤지 -0.871926
BGF -0.840062
CJ대한통운 -0.820994
한국전력 -0.798898
하이트론 -0.793636
Name: 구리, dtype: float64
col = ['미래아이앤지', 'BGF', 'CJ대한통운', '한국전력', '하이트론', '구리']
x = FTD[col].values
min_max_scaler = preprocessing.MinMaxScaler()
x_scaled = min_max_scaler.fit_transform(x.astype(float))
cu = pd.DataFrame(x_scaled, columns = col, index = FTD.index)
cu.head()
| 미래아이앤지 | BGF | CJ대한통운 | 한국전력 | 하이트론 | 구리 | |
|---|---|---|---|---|---|---|
| date | ||||||
| 2015-01-07 | 0.401904 | 0.306223 | 0.547739 | 0.483740 | 0.279188 | 0.644206 |
| 2015-01-19 | 0.381209 | 0.308971 | 0.532663 | 0.379404 | 0.231810 | 0.471332 |
| 2015-02-03 | 0.401904 | 0.354307 | 0.713568 | 0.460705 | 0.200508 | 0.479647 |
| 2015-02-12 | 0.354305 | 0.368045 | 0.683417 | 0.531165 | 0.370558 | 0.479647 |
| 2015-02-27 | 0.377070 | 0.376288 | 0.522613 | 0.525745 | 0.494078 | 0.543911 |
plt.plot(cu['미래아이앤지'],ls = ":", label = '미래아이앤지')
plt.plot(cu['BGF'],ls = ":", label = 'BGF')
plt.plot(cu['CJ대한통운'],ls = ":", label = 'CJ대한통운')
plt.plot(cu['한국전력'],ls = ":", label = '한국전력')
plt.plot(cu['하이트론'],ls = ":", label = '하이트론')
plt.plot(cu['구리'], label = '구리')
plt.legend(loc = 'best')
plt.grid()
plt.show()
corr_TD.iloc[:,1].sort_values(ascending=True).head()
itemname
BYC -0.844172
NPC -0.827650
동성화학 -0.823397
조일알미늄 -0.821647
미래아이앤지 -0.814216
Name: 난방유, dtype: float64
col = ['BYC', 'NPC', '동성화학', '조일알미늄', '미래아이앤지', '난방유']
x = FTD[col].values
min_max_scaler = preprocessing.MinMaxScaler()
x_scaled = min_max_scaler.fit_transform(x.astype(float))
cu = pd.DataFrame(x_scaled, columns = col, index = FTD.index)
plt.plot(cu['BYC'],ls = ":", label = 'BYC')
plt.plot(cu['NPC'],ls = ":", label = 'NPC')
plt.plot(cu['동성화학'],ls = ":", label = '동성화학')
plt.plot(cu['조일알미늄'],ls = ":", label = '조일알미늄')
plt.plot(cu['미래아이앤지'],ls = ":", label = '미래아이앤지')
plt.plot(cu['난방유'], label = '난방유')
plt.legend(loc = 'best')
plt.grid()
plt.show()
corr_TD.iloc[:,13].sort_values(ascending=True).head()
itemname
한세예스24홀딩스 -0.901462
쿠쿠홀딩스 -0.885435
S&T홀딩스 -0.884960
전방 -0.870637
사조해표 -0.868335
Name: 주석, dtype: float64
col = ['한세예스24홀딩스', '쿠쿠홀딩스', 'S&T홀딩스', '전방', '사조해표', '주석']
x = FTD[col].values
min_max_scaler = preprocessing.MinMaxScaler()
x_scaled = min_max_scaler.fit_transform(x.astype(float))
cu = pd.DataFrame(x_scaled, columns = col, index = FTD.index)
plt.plot(cu['한세예스24홀딩스'],ls = ":", label = '한세예스24홀딩스')
plt.plot(cu['쿠쿠홀딩스'],ls = ":", label = '쿠쿠홀딩스')
plt.plot(cu['S&T홀딩스'],ls = ":", label = 'S&T홀딩스')
plt.plot(cu['전방'],ls = ":", label = '전방')
plt.plot(cu['사조해표'],ls = ":", label = '사조해표')
plt.plot(cu['주석'], label = '주석')
plt.legend(loc = 'best')
plt.grid()
plt.show()
