[비트코인 전처리2] 트윗 데이터에 가격 태그 붙이기

 

 

'''Adding pirce features to twit data'''
 
#구글 드라이브 마운트
 
from google.colab import drive
drive.mount('/content/drive')
 
 
# 라이브러리 임포트
 
import pandas as pd
import numpy as np
from tqdm import tqdm 
 
#데이터 임포트, temp는 100개 선별 데이터.
 
temp = pd.read_csv("/content/drive/My Drive/BIGCOIN/PREPORCOESSING/트레인데이터/temp_100_selected.csv")
bitstamp = pd.read_csv("/content/drive/My Drive/BIGCOIN/PREPORCOESSING/bitcoin-historical-data/bitstampUSD_1-min_data_2012-01-01_to_2019-08-12.csv")
coinbase = pd.read_csv("/content/drive/My Drive/BIGCOIN/PREPORCOESSING/bitcoin-historical-data/coinbaseUSD_1-min_data_2014-12-01_to_2019-01-09.csv")
 
 
#데이터 결측치 제거 및 인덱스 초기화
 
 
bitstamp_na = bitstamp.dropna(axis=0)
bitstamp_dist=bitstamp_na.drop_duplicates()
bitstamp_reset= bitstamp_dist.reset_index(drop=True)
 
 
coinbase_na = coinbase.dropna(axis=0)
coinbase_dist=coinbase_na.drop_duplicates()
coinbase_reset= coinbase_dist.reset_index(drop=True) 
 
 
# 가격데이터와, 트레인 데이터 조인으로 필요한 데이터만 선택
 
 
temp = temp[["text", "hour_stamp"]]
bitstamp_reset=bitstamp_reset.rename(columns = {'Timestamp':'hour_stamp'})
coinbase_reset=coinbase_reset.rename(columns = {'Timestamp':'hour_stamp'})
 
merge_temp_bit = pd.merge(temp, bitstamp_reset, how = 'left', on ='hour_stamp')
merge_temp_co = pd.merge(temp, coinbase_reset, how = 'left', on ='hour_stamp')
 
 
 
#결측치가 bitstamp가 더 적음, bitstamp 선택
 
print(merge_temp_co.isnull().sum())
print(merge_temp_bit.isnull().sum())
 
 
# 본래의 bitstamp열만 분리
bitstamp_set =merge_temp_bit[["hour_stamp","Open","High","Low",'Close','Volume_(BTC)','Volume_(Currency)', 'Weighted_Price']]
 
 
#중복 제거
bitstamp_set=bitstamp_set.drop_duplicates()
bitstamp_set= bitstamp_set.reset_index(drop=True)
 
 
 
 
 
# 차잇값 계산 및 추가
#for i in tqdm(range(len(bitstamp)) leave=False):
gap_Open =[bitstamp_set["Open"][i]-bitstamp_set["Open"][i-1] for i in tqdm(range(1, len(bitstamp_set)), miniters=500000)]
  #gap_Open.append(bitstamp["Open"][i]-bitstamp["Open"][i-1])
 
gap_High =[bitstamp_set["High"][i]-bitstamp_set["High"][i-1] for i in tqdm(range(1, len(bitstamp_set)),miniters=500000)]
  #gap_High.append(bitstamp["High"][i]-bitstamp["High"][i-1])
  
gap_Low =[bitstamp_set["Low"][i]-bitstamp_set["Low"][i-1] for i in tqdm(range(1, len(bitstamp_set)),miniters=500000)]
  #gap_Low.append(bitstamp["Low"][i]-bitstamp["Low"][i-1])
 
gap_vol_BTC =[bitstamp_set["Volume_(BTC)"][i]-bitstamp_set["Volume_(BTC)"][i-1] for i in tqdm(range(1, len(bitstamp_set)),miniters=500000)]
  #gap_vol_BTC.append(bitstamp["Volume_(BTC)"][i]-bitstamp["Volume_(BTC)"][i-1])
 
gap_vol_Currency =[bitstamp_set["Volume_(Currency)"][i]-bitstamp_set["Volume_(Currency)"][i-1] for i in tqdm(range(1, len(bitstamp_set)),miniters=500000)]
  #gap_vol_Currency.append(bitstamp["Volume_(Currency)"][i]-bitstamp["Volume_(Currency)"][i-1])
 
gap_weighted_price =[bitstamp_set["Weighted_Price"][i]-bitstamp_set["Weighted_Price"][i-1] for i in tqdm(range(1, len(bitstamp_set)),miniters=500000)]
  #gap_weighted_price .append(bitstamp["Weighted_Price"][i]-bitstamp["Weighted_Price"][i-1])
 
# 첫째 원소 추가, 첫째 값은 계산이 불가능
 
gap_Open.insert(0,0.55)
gap_High.insert(0, 1.1)
gap_Low.insert(0, 0.55)
gap_vol_BTC.insert(0,8.145574)
gap_vol_Currency.insert(0,7066.100435)
gap_weighted_price.insert(0,0.736157)
 
 
 
#bitstamp_set 데이터프레임에 열 추가
bitstamp_set["gap_Open"] = gap_Open
bitstamp_set["gap_High"] = gap_High
bitstamp_set["gap_Low"] = gap_Low
bitstamp_set["gap_vol_BTC"] = gap_vol_BTC
bitstamp_set["gap_vol_Currency"] = gap_vol_Currency
bitstamp_set["gap_weighted_price"] = gap_weighted_price
 
 
 
# 부호 계산 및 열 추가
 
  # BTC Open 부호
 #for i in range(len(bitstamp)):
 
 sign_gap_Open =[-1 if bitstamp_set["gap_Open"][i] < 0  else 1 if bitstamp_set["gap_Open"][i] > 0 else 0 for i in tqdm(range(len(bitstamp_set)))]
 
 ''' #if bitstamp["gap_Open"][i] < 0:
    sign_gap_Open.append(-1)
  elif bitstamp["gap_Open"][i] > 0:
   sign_gap_Open.append(1)
  elif bitstamp["gap_Open"][i] == 0:
    sign_gap_Open.append(0)
'''
 
 
    # BTC High 부호
 '''   
  if bitstamp["gap_High"][i] < 0:
    sign_gap_High.append(-1)
  elif bitstamp["gap_High"][i] > 0:
    sign_gap_High.append(1)
  elif bitstamp["gap_High"][i] == 0:
    sign_gap_High.append(0)
'''
 
 sign_gap_High =[-1 if bitstamp_set["gap_High"][i] < 0  else 1 if bitstamp_set["gap_High"][i] > 0 else 0 for i in tqdm(range(len(bitstamp_set)))]
 
 
    # BTC Low 부호
'''    
  if bitstamp["gap_Low"][i] < 0:
    sign_gap_Low.append(-1)
  elif bitstamp["gap_Low"][i] > 0:
    sign_gap_Low.append(1)
  elif bitstamp["gap_Low"][i] == 0:
    sign_gap_Low.append(0)
'''
sign_gap_Low =[-1 if bitstamp_set["gap_Low"][i] < 0  else 1 if bitstamp_set["gap_Low"][i] > 0 else 0 for i in tqdm(range(len(bitstamp_set)))]
 
 
 
  # BTC 거래량 부호
 ''' 
  if bitstamp["Volume_(BTC)"][i] < 0:
    sign_vol_BTC.append(-1)
  elif bitstamp["Volume_(BTC)"][i] > 0:
    sign_vol_BTC.append(1)
  elif bitstamp["Volume_(BTC)"][i] == 0:
    sign_vol_BTC.append(0)
  '''
sign_vol_BTC =[-1 if bitstamp_set["Volume_(BTC)"][i] < 0  else 1 if bitstamp_set["Volume_(BTC)"][i] > 0 else 0 for i in tqdm(range(len(bitstamp_set)))]
 
 
  #통화 거래량 부호
 ''' 
  if bitstamp["gap_vol_Currency"][i] < 0:
    sign_vol_currency.append(-1)
  elif bitstamp["gap_vol_Currency"][i] > 0:
    sign_vol_currency.append(1)
  elif bitstamp["gap_vol_Currency"][i] == 0:
    sign_vol_currency.append(0)  
  '''
sign_vol_currency =[-1 if bitstamp_set["gap_vol_Currency"][i] < 0  else 1 if bitstamp_set["gap_vol_Currency"][i] > 0 else 0 for i in tqdm(range(len(bitstamp_set)))]
 
  #가격 변화량 부호
 ''' 
  if bitstamp["Weighted_Price"][i] < 0:
    sign_vol_weighted_price.append(-1)
  elif bitstamp["Weighted_Price"][i] > 0:
    sign_vol_weighted_price.append(1)
  elif bitstamp["Weighted_Price"][i] == 0:
    sign_vol_weighted_price.append(0)  
 '''
 
 sign_vol_weighted_price =[-1 if bitstamp_set["Weighted_Price"][i] < 0  else 1 if bitstamp_set["Weighted_Price"][i] > 0 else 0 for i in tqdm(range(len(bitstamp_set)))]
 
 
# bitstamp 데이터 프레임에 열 추가
bitstamp_set["sign_gap_Open"] = sign_gap_Open
bitstamp_set["sign_gap_High"] = sign_gap_High
bitstamp_set["sign_gap_Low"] = sign_gap_Low
bitstamp_set["sign_vol_BTC"] = sign_vol_BTC
bitstamp_set["sign_vol_currency"] = sign_vol_currency
bitstamp_set["sign_vol_weighted_price"] = sign_vol_weighted_price
 
# train 필요한 열만 선택
bitstamp_set.head()
 
# 다시 조인해서 데이터 붙여주기
merge_temp_bit = pd.merge(temp, bitstamp_set, how = 'left', on ='hour_stamp')
 
 
# 0값으로 있는 중립치에 대해서 제거
 
merge_temp_bit = merge_temp_bit.dropna(axis=0)
merge_temp_bit= merge_temp_bit.reset_index(drop=True)
 
 
# 결측치 확인
print(merge_temp_bit.isnull().sum())
 
 
#저장
merge_temp_bit.to_csv("merge_temp_bit(complete).csv")