import csv
import requests
import datetime
import pandas as pd
import seaborn
import matplotlib.pyplot as plt
import numpy as np
import math

from sklearn.linear_model import LinearRegression
from sklearn import preprocessing, cross_validation, svm

# Pretty print the JSON
import uuid
from IPython.display import display_javascript, display_html, display
import json

class RenderJSON(object):
    def __init__(self, json_data):
        if isinstance(json_data, dict):
            self.json_str = json.dumps(json_data)
        else:
            self.json_str = json_data
        self.uuid = str(uuid.uuid4())

    def _ipython_display_(self):
        display_html('<div id="{}" style="height: 600px; width:100%;"></div>'.format(self.uuid), raw=True)
        display_javascript("""
        require(["https://rawgit.com/caldwell/renderjson/master/renderjson.js"], function() {
        document.getElementById('%s').appendChild(renderjson(%s))
        });
        """ % (self.uuid, self.json_str), raw=True)

#Defining Price Data From Crypto cryptocompare.com
def price(symbol, comparison_symbols=['BTC'], exchange=''):
    url = 'https://min-api.cryptocompare.com/data/price?fsym={}&tsyms={}'\
            .format(symbol.upper(), ','.join(comparison_symbols).upper())
    if exchange:
        url += '&e={}'.format(exchange)
    page = requests.get(url)
    data = page.json()
    return data

# Displaying Currency
price('LTC', exchange='binance')

{'BTC': 0.01268}

# Pulling Historical Minute Data From cryptocompare.com API
def minute_price_historical(symbol, comparison_symbol, limit, aggregate, exchange=''):
    url = 'https://min-api.cryptocompare.com/data/histominute?fsym={}&tsym={}&limit={}&aggregate={}'\
            .format(symbol.upper(), comparison_symbol.upper(), limit, aggregate)
    if exchange:
        url += '&e={}'.format(exchange)
    page = requests.get(url)
    data = page.json()['Data']
    df = pd.DataFrame(data)
    df['timestamp'] = [datetime.datetime.fromtimestamp(d) for d in df.time]
    return df

time_delta = 1 # Bar width in minutes
df = minute_price_historical('BTC', 'USD', 9999, time_delta)
df = df[['close','high','low','open','time','volumefrom','volumeto']]
#print('Max length = %s' % len(df))
#print('Max time = %s' % (df.timestamp.max() - df.timestamp.min()))

time_delta = 1 # Bar width in minutes
ts = minute_price_historical('BTC', 'USD', 9999, time_delta)
ts = ts[['close','high','low','open','time','volumefrom','volumeto','timestamp']]
#print('Max length = %s' % len(df))
#print('Max time = %s' % (df.timestamp.max() - df.timestamp.min()))

# Plotting Pulled Data
plt.title('BTC vs USD')
plt.xlabel('Time')
plt.ylabel('Price')
plt.plot(ts.timestamp, df.close)
plt.xticks(rotation=45)
plt.show()
#df.to_csv('btc_minute.csv')

# Displaying Data
df.tail()

#forecast_out = int(30) # predicting 21 minutes into future
forecast_out = int(math.ceil(0.01*len(df)))
df['Prediction'] = df[['close']].shift(-forecast_out) #  label column with data shifted 21 units up
#df.to_csv('prediction')

X = np.array(df.drop(['Prediction'], 1))
X = preprocessing.scale(X)

X_forecast = X[-forecast_out:] # set X_forecast equal to last 21
X = X[:-forecast_out] # remove last 21 from X

y = np.array(df['Prediction'])
y = y[:-forecast_out]

X_train, X_test, y_train, y_test = cross_validation.train_test_split(X, y, test_size = 0.2)

# Training
clf = LinearRegression()
clf.fit(X_train,y_train)
# Testing
confidence = clf.score(X_test, y_test)
print("confidence: ", confidence)

confidence:  0.962726183878

forecast_prediction = clf.predict(X_forecast)
forecast_prediction

array([ 6354.81457621,  6354.84693344,  6355.51358134,  6355.5441647 ,
        6354.10256002,  6352.1649325 ,  6350.36762175,  6349.71327774,
        6346.20948474,  6343.13710109,  6347.02449051,  6350.73222549,
        6351.94186358,  6351.87095975,  6353.19890519,  6353.49974969,
        6352.05390689,  6352.81579194,  6353.14852856,  6353.29820626,
        6352.74883958])

pre = pd.DataFrame(forecast_prediction)
pre.columns = ['prediction']

# Date magic for our dataframe 
now = datetime.datetime.now()

time = pd.date_range(now,periods=21, freq='1min')
time
minute = pd.DataFrame(time)
#start = pre.index.min().date() 
#end = pre.index.max().date() + pd.Timedelta(1, 'D')
#pre.reindex(pd.date_range(start, end, freq='T', closed='left')).fillna(method='ffill')

pre.tail()
#date = pd.date_range(start='4/1/2018', end='4/1/2018', freq='m')
#date = pd.DataFrame(date)
#date.columns = ['date']
#date

minute.tail()

prediction = pd.concat([pre,minute], axis=1)
prediction.columns = ['prediction','time']

sell = prediction.prediction.max()
prediction.tail()

#buy = prediction.prediction.min()

#signal =(buy,sell)
#signal
#stuff = pd.concat([buy,sell], axis=1)
#stuff

#gains = sell/buy
#gains

confidence

0.96272618387805275

plt.suptitle('21 Minute Future Prediction BTC vs USD',fontsize=17)
#plt.title('Potential Gains %s' % gains, fontsize=10)
plt.xlabel('Time')
plt.ylabel('Price')
plt.plot(prediction.time, prediction.prediction)
plt.xticks(rotation=45)
plt.show()

time_delta = 1 # Line width in minutes
ts = minute_price_historical('BTC', 'USD', 9999, time_delta)
ts = ts[['close','high','low','open','time','volumefrom','volumeto','timestamp']]

ts = ts.tail(19)

plt.title('Past BTC vs USD Data', fontsize=17)
plt.xlabel('Time')
plt.ylabel('Price')
plt.plot(ts.timestamp, ts.close)
plt.xticks(rotation=45)
plt.show()

	close	high	low	open	time	volumefrom	volumeto
1996	6344.39	6346.10	6341.66	6341.69	1530381960	21.59	137489.93
1997	6342.10	6347.01	6341.95	6344.39	1530382020	24.00	152175.73
1998	6344.86	6347.16	6342.07	6342.07	1530382080	25.57	162427.74
1999	6346.48	6347.11	6342.81	6344.86	1530382140	23.02	146362.87
2000	6344.77	6346.86	6344.77	6346.47	1530382200	8.25	52359.32

	prediction	time
16	6352.053907	2018-06-30 14:26:59.090775
17	6352.815792	2018-06-30 14:27:59.090775
18	6353.148529	2018-06-30 14:28:59.090775
19	6353.298206	2018-06-30 14:29:59.090775
20	6352.748840	2018-06-30 14:30:59.090775