def SLR(df, price='Close', predictions_int=0):
"""
Simple Linear Regression
Returns: list of floats = jhta.SLR(df, price='Close', predictions_int=0)
Source: https://machinelearningmastery.com/implement-simple-linear-regression-scratch-python/
"""
x_list = list(range(len(df[price]) - predictions_int))
p_list = df[price][0:len(df[price]) - predictions_int]
b1 = jhta.COV(x_list, p_list) / jhta.VARIANCE({'x': x_list}, len(x_list),
'x')[-1]
b0 = jhta.MEAN({'y': p_list}, len(p_list), 'y')[-1] - b1 * jhta.MEAN(
{'x': x_list}, len(x_list), 'x')[-1]
return [b0 + b1 * i for i in range(len(df[price]))]
def COV(x_list, y_list):
"""
Covariance
Returns: float = jhta.COV(x_list, y_list)
Source: https://en.wikipedia.org/wiki/Algorithms_for_calculating_variance#Covariance
"""
x_mean = jhta.MEAN({'x_list': x_list}, len(x_list), 'x_list')[-1]
y_mean = jhta.MEAN({'y_list': y_list}, len(y_list), 'y_list')[-1]
covariance = .0
for i in range(len(x_list)):
a = x_list[i] - x_mean
b = y_list[i] - y_mean
covariance += a * b / len(x_list)
return covariance
def STANDARDIZE(df, price='Close'):
"""
Standardize
Returns: list of floats = jhta.STANDARDIZE(df, price='Close')
Source: https://machinelearningmastery.com/normalize-standardize-time-series-data-python/
"""
standardize_list = []
start = None
for i in range(len(df[price])):
if df[price][i] != df[price][i] or i < 1:
standardize = float('NaN')
else:
if start is None:
start = i
x = df[price][start:]
mean = jhta.MEAN({'x': x}, len(x), 'x')[-1]
standard_deviation = jhta.STDEV({'x': x}, len(x), 'x')[-1]
standardize = (df[price][i] - mean) / standard_deviation
standardize_list.append(standardize)
return standardize_list
def STANDARDIZE(df, price='Close'):
"""
Standardize
"""
standardize_list = []
i = 0
start = None
while i < len(df[price]):
if df[price][i] != df[price][i] or i < 1:
standardize = float('NaN')
else:
if start is None:
start = i
x = df[price][start:]
mean = jhta.MEAN({'x': x}, len(x), 'x')[-1]
standard_deviation = jhta.STDEV({'x': x}, len(x), 'x')[-1]
standardize = (df[price][i] - mean) / standard_deviation
standardize_list.append(standardize)
i += 1
return standardize_list
def INFO(df, price='Close'):
"""
Print df Information
"""
print('{:_<28}:{:_>22}'.format('DF PRICE COLUMN', price))
print('{:_<28}:{:_>22d}'.format('LEN', len(df[price])))
print('{:_<28}:{:_>28.5f}'.format('MIN',
jhta.MIN(df, len(df[price]), price)[-1]))
print('{:_<28}:{:_>28.5f}'.format('MAX',
jhta.MAX(df, len(df[price]), price)[-1]))
print('{:_<28}:{:_>28.5f}'.format('SUM',
jhta.SUM(df, len(df[price]), price)[-1]))
print('{:_<28}:{:_>28.5f}'.format('MEAN',
jhta.MEAN(df, len(df[price]),
price)[-1]))
# print ('{:_<28}:{:_>28.5f}'.format('HARMONIC_MEAN', jhta.HARMONIC_MEAN(df, len(df[price]), price)[-1]))
print('{:_<28}:{:_>28.5f}'.format(
'MEDIAN',
jhta.MEDIAN(df, len(df[price]), price)[-1]))
print('{:_<28}:{:_>28.5f}'.format(
'MEDIAN_LOW',
jhta.MEDIAN_LOW(df, len(df[price]), price)[-1]))
print('{:_<28}:{:_>28.5f}'.format(
'MEDIAN_HIGH',
jhta.MEDIAN_HIGH(df, len(df[price]), price)[-1]))
print('{:_<28}:{:_>28.5f}'.format(
'MEDIAN_GROUPED',
jhta.MEDIAN_GROUPED(df, len(df[price]), price)[-1]))
# print ('{:_<28}:{:_>28.5f}'.format('MODE', jhta.MODE(df, len(df[price]), price)[-1]))
print('{:_<28}:{:_>28.5f}'.format(
'PSTDEV',
jhta.PSTDEV(df, len(df[price]), price)[-1]))
print('{:_<28}:{:_>28.5f}'.format(
'PVARIANCE',
jhta.PVARIANCE(df, len(df[price]), price)[-1]))
print('{:_<28}:{:_>28.5f}'.format(
'STDEV',
jhta.STDEV(df, len(df[price]), price)[-1]))
print('{:_<28}:{:_>28.5f}'.format(
'VARIANCE',
jhta.VARIANCE(df, len(df[price]), price)[-1]))
def main():
# quandl_data = quandl.get('BCHARTS/BITSTAMPUSD', order='asc', collapse='daily', returns='numpy', authtoken='YOUR_AUTH_TOKEN')
quandl_data = quandl.get('BCHARTS/BITSTAMPUSD', order='asc', collapse='daily', returns='numpy')
df = {'datetime': [], 'Open': [], 'High': [], 'Low': [], 'Close': [], 'Volume': []}
i = 0
while i < len(quandl_data['Close']):
# df['datetime'].append(i)
df['datetime'].append(quandl_data['Date'][i])
df['Open'].append(float(quandl_data['Open'][i]))
df['High'].append(float(quandl_data['High'][i]))
df['Low'].append(float(quandl_data['Low'][i]))
df['Close'].append(float(quandl_data['Close'][i]))
df['Volume'].append(int(quandl_data['Volume (BTC)'][i]))
i += 1
x = df['datetime']
sma_list = jhta.SMA(df, 200)
mmr_list = jhta.MMR(df)
mmr_mean_list = jhta.MEAN({'mmr': mmr_list}, len(mmr_list), 'mmr')
mom_list = jhta.MOM(df, 365)
mom_mean_list = jhta.MEAN({'mom': mom_list}, len(mom_list), 'mom')
print ('Calculated from %i data points:' % len(x))
print ('Last Close: %f' % df['Close'][-1])
print ('Last SMA 200: %f' % sma_list[-1])
print ('Last MMR: %f' % mmr_list[-1])
print ('Last MEAN MMR: %f' % mmr_mean_list[-1])
print ('Last MOM 365: %f' % mom_list[-1])
print ('Last MEAN MOM 365: %f' % mom_mean_list[-1])
# left = 365
# right = len(x)
# print ('Plot starts from %i until %i in Log scale:' % (left, right))
plt.figure(1, (30, 10))
plt.subplot(311)
plt.title('Time / Price / Ratio')
plt.xlabel('Time')
plt.ylabel('Price')
plt.grid(True)
plt.plot(x, df['Close'], color='blue')
plt.plot(x, sma_list, color='red')
plt.legend(['Close', 'SMA 200'], loc='upper left')
plt.gca().xaxis.set_major_formatter(mdates.DateFormatter('%Y-%m-%d'))
plt.gca().xaxis.set_major_locator(mdates.YearLocator())
plt.gcf().autofmt_xdate()
# plt.xlim(left=left, right=right)
plt.yscale('log')
plt.subplot(312)
plt.xlabel('Time')
plt.ylabel('Ratio')
plt.grid(True)
plt.plot(x, [1] * len(x), color='red')
plt.plot(x, mmr_list)
plt.plot(x, mmr_mean_list)
plt.plot(x, [2.4] * len(x))
plt.legend(['SMA 200', 'MMR', 'MEAN MMR', 'THRESHOLD 2.4'], loc='upper left')
plt.gca().xaxis.set_major_formatter(mdates.DateFormatter('%Y-%m-%d'))
plt.gca().xaxis.set_major_locator(mdates.YearLocator())
plt.gcf().autofmt_xdate()
# plt.xlim(left=left, right=right)
plt.yscale('log')
plt.subplot(313)
plt.xlabel('Time')
plt.ylabel('Ratio')
plt.grid(True)
plt.plot(x, [0] * len(x), color='blue')
plt.plot(x, mom_list)
plt.plot(x, mom_mean_list)
plt.legend(['Price', 'MOM 365', 'MEAN MOM 365'], loc='upper left')
plt.gca().xaxis.set_major_formatter(mdates.DateFormatter('%Y-%m-%d'))
plt.gca().xaxis.set_major_locator(mdates.YearLocator())
plt.gcf().autofmt_xdate()
# plt.xlim(left=left, right=right)
plt.yscale('symlog')
plt.show()