def merge_lstm_data(starttime='2016-06-24',
split_date='2018-06-20',
period=900,
features=[
'_volatility', '_date', '_close', '_len_day_data',
'_neg_res', '_neut_tweet', '_pos_tweet'
]):
filestr = 'start_' + str(starttime) + '_split_' + str(
split_date) + '_period_' + str(period)
try:
training_set = pd.read_csv('CurDat/temp_results_train_' + filestr +
'.csv')
test_set = pd.read_csv('CurDat/temp_results_train_' + filestr + '.csv')
model_data = pd.read_csv('CurDat/temp_results_model_' + filestr +
'.csv')
print('Loaded data from CurDat/temp_results_train_' + filestr +
'.csv!')
print('Loaded data from CurDat/temp_results_test_' + filestr + '.csv!')
print('Loaded data from CurDat/temp_results_model_' + filestr +
'.csv!')
except:
####### important parameters##################
print("Merge coin & twit data sets..")
selected_coins = ['BTC', 'ETH'] #['BTC', 'LTC', 'ETH', 'XMR']
# period = 900#86400 candlestick period in seconds; valid values are 300, 900, 1800, 7200, 14400, and 86400),
# split_date = '2018-06-20'
# window_len = 25
#######################################
df_ini = data_reader.get_poloniex_data(startdate=starttime)
coin_list = df_ini.fetch_coin_data(selected_coins, period)
import raspi_tweet_fetcher.Load_Tweets_Class as LTC
### init
get_tweet_data = LTC.get_tweets()
### fetch data based on query word
#get_tweet_data.fetch_tweets(query='BITCOIN', count=100, pages=1)
#get_tweet_data.fetch_stocktwits(query='BITCOIN')
### delete duplicates data from CSV
tweet_data = get_tweet_data.read_and_clean_data_from_csv(
query='BITCOIN')
#data = get_tweet_data.data
fin_data = coin_list['BTC']
### analyze sentiment
tweet_results = get_tweet_data.analyze_Tweets(tweet_data)
fin_data['Date'] = pd.to_datetime(fin_data.date_format)
tweet_results['Date'] = pd.to_datetime(tweet_results['days'])
merged = pd.merge(fin_data, tweet_results, how='outer', on='Date')
merged.dropna(how='any', inplace=True)
new_coin_list = {}
new_coin_list['BTC'] = merged
training_set, test_set, model_data = create_coin_dataset(
new_coin_list,
starttime=starttime,
split_date=split_date,
features=[
'_volatility', '_date', '_close', '_len_day_data', '_neg_res',
'_neut_tweet', '_pos_tweet'
]) # '_date','_close_off_high'])
#
# features = ['_close','_volume','_volatility','_date', '_weightedAverage' ,
# '_len_day_data' , '_neg_res', '_neut_tweet', '_pos_tweet'])
panda_train = pd.DataFrame(training_set)
panda_train.to_csv('CurDat/temp_results_train_' + filestr + '.csv')
panda_test = pd.DataFrame(test_set)
panda_test.to_csv('CurDat/temp_results_test_' + filestr + '.csv')
panda_model = pd.DataFrame(model_data)
panda_model.to_csv('CurDat/temp_results_model_' + filestr + '.csv')
return training_set, test_set, model_data
@author: Chris
"""
import data_reader as data_reader
import raspi_tweet_fetcher.Load_Tweets_Class as LTC
from textblob import TextBlob
import re
import os
import pandas as pd
selected_coins = ['BTC', 'ETH'] #['BTC', 'LTC', 'ETH', 'XMR']
period = 7200#86400
split_date = '2018-06-15'
df_ini = data_reader.get_poloniex_data()
coin_list = df_ini.fetch_coin_data(selected_coins, period)
### init
get_tweet_data = LTC.get_tweets()
### fetch data based on query word
#get_tweet_data.fetch_tweets(query='BITCOIN', count=100, pages=1)
#get_tweet_data.fetch_stocktwits(query='BTC.X')
### delete duplicates data from CSV