[주식 - NC SOFT] 주식 데이터 예측하기

NC SOFT 주식 데이터를 이용하여 2년간 주식을 예측해보았다

 

주식데이터 분석하기¶

 

• 시: 시작가
• 고: 고가
• 저: 최저가
• 종: 종가(해당시점) / 수정종가

In [2]:

from pandas_datareader import data
import datetime

 

야후 fianace이용 :https://finance.yahoo.com/¶

In [1]:

import fix_yahoo_finance as yf
yf.pdr_override()

In [4]:

start_date = '2008-01-01'
name = '036570.KS'
nc = data.get_data_yahoo(name, start_date)

 

[*********************100%***********************]  1 of 1 downloaded

 

ns soft 주식을 볼 것이당¶

In [6]:

nc.head(3)

Out[6]:

	Open	High	Low	Close	Adj Close	Volume
Date
2008-01-02	49000.0	50100.0	48600.0	48700.0	43573.867188	83601
2008-01-03	48650.0	49500.0	48500.0	49150.0	43976.496094	49680
2008-01-04	48700.0	49200.0	48000.0	48750.0	43618.597656	88832

In [12]:

import pandas as pd
import matplotlib.pyplot as plt
%matplotlib inline

In [9]:

nc['Close'].plot(figsize=(12, 6), grid=True)

Out[9]:

<matplotlib.axes._subplots.AxesSubplot at 0x10d694e10>

 

In [11]:

nc_trunc = nc[:'2016-12-31']
nc_trunc.head(3)

Out[11]:

	Open	High	Low	Close	Adj Close	Volume
Date
2008-01-02	49000.0	50100.0	48600.0	48700.0	43573.867188	83601
2008-01-03	48650.0	49500.0	48500.0	49150.0	43976.496094	49680
2008-01-04	48700.0	49200.0	48000.0	48750.0	43618.597656	88832

In [13]:

df = pd.DataFrame({'ds':nc_trunc.index, 'y':nc_trunc['Close']})
df.reset_index(inplace=True)
del df['Date']
df.head(3)

Out[13]:

	ds	y
0	2008-01-02	48700.0
1	2008-01-03	49150.0
2	2008-01-04	48750.0

In [14]:

from fbprophet import Prophet

In [26]:

m = Prophet()
m.fit(df);

 

 

In [27]:

future = m.make_future_dataframe(periods=365*2)
future.tail(3)

Out[27]:

	ds
2962	2018-12-27
2963	2018-12-28
2964	2018-12-29

In [28]:

forecast = m.predict(future)
forecast[['ds', 'yhat', 'yhat_lower', 'yhat_upper']].tail(3)

Out[28]:

	ds	yhat	yhat_lower	yhat_upper
2962	2018-12-27	357156.014083	125914.870456	584369.690974
2963	2018-12-28	356983.370256	133279.020838	585731.617123
2964	2018-12-29	355064.844669	128056.489530	587925.729847

In [29]:

m.plot(forecast)

Out[29]:

 

In [30]:

m.plot_components(forecast)

Out[30]:

 

 

이제 2017년도 기록을 가지고 실제데이터와 예측데이터를 비교해보자¶

In [31]:

plt.figure(figsize=(12, 6))
plt.plot(nc.index, nc['Close'], label='real')
plt.plot(forecast['ds'], forecast['yhat'], label='forecast')
plt.grid()
plt.legend()
plt.show()