Leveling-ed-StockData.py

#True

# Level data so series is stationary in time
# take log of data
# save it to use in deconstructing signal to find anomolies
# Using finance.yahoo.com Nasdaq, S&P, DJI 1985 - date (Nov 29 2017)
#

import numpy as np
import pandas as pd
import matplotlib.pyplot as plt


# pandas display options
pd.options.display.max_rows = 10000
pd.options.display.max_columns = 25
pd.options.display.width = 1000



######################################################################
# plot dataframe
########################################################################
def plot_dataframe(d, t):


    plt.figure(figsize=(18,18))
    plt.plot(d['NASDAQ'], label='NASDAQ')
    plt.plot(d['S&P'], label='S&P')
    plt.plot(d['DJIA'], label='DJIA')
    plt.plot(d['BTC'], label='BTC')
    plt.plot(d['Russell'], label='Russell')
    plt.title(t)
    plt.legend(loc='best')
    plt.show()


######################################################################
# data
########################################################################
# read in datafile created in LoadAndMatchDates.py
data = pd.read_csv('./StockDataWithVolume.csv', index_col='Date', parse_dates=True)
features = ['DJIA', 'S&P', 'NASDAQ', 'Russell', 'BTC']

# fill in a couple NaN
# data.dropna()
data = data.fillna(method='ffill')


#########################################################################################
# level the series out, time series calculations all assume signal is stationary in time
########################################################################################

# pandas removed ols package !#&^*@$
# need y intercept, b
# and slope, m
# y = mx + b
# using simplest case possible
#
# how to get x, y just in case you want to put this into an ordinary least squares package
# for better slope/intercept numbers
# This is close enough for proof of concept
#
# x = list(range(1, len(data)))
# y = data


# son terim - ilk terim bolu kolon uzunlugu egim ve ilk terimi de constant b olarak belirledik.
# not really ols, but close enough
def ols(data):
    m = (data[-1] - data[0]) / len(data)
    b = data[0]

    print(data[-1], data[0], (data[-1] - data[0]))
    print(m, b)

    print('-----------------------')

    return m, b


# add a time step
steps = np.asarray(range(1, len(data)+1))
print('steps')
print(steps)
steps.reshape(1, -1)
data['step'] = steps