tools_predict_generic.py

import os, sys

try:

    import time

except:

    os.system('pip install time')

    import time

try:

    import math

except:

    os.system('pip install math')

    import math

try:

    import csv

except:

    os.system("pip3 install csv")

    import csv

try:

    import requests

except:

    os.system('pip install requests')

    import requests

try:

    import matplotlib as mpl

except:

    os.system('pip3 install matplotlib')

    import matplotlib as mpl

try:

    import matplotlib.pyplot as plt

except:

    os.system("pip3 install matplotlib")

    import matplotlib.pyplot as plt

from  matplotlib import style

try:

    import pandas as pd

except:

    os.system('pip3 install pandas')

    import pandas as pd

try:

    import numpy as np

except:

    os.system("pip3 install numpy")

    import numpy as np

try:

    from sklearn.svm import SVR

except:

    os.system('pip install sklearn.svm')

    from sklearn.svm import SVR

try:

    from sklearn.model_selection import train_test_split

except:

    os.system('pip install sklearn.model_selection')

    from sklearn.model_selection import train_test_split

try:

    from sklearn.svm import SVR

except:

    os.system('pip install sklearn.svm')

    from sklearn.svm import SVR

try:

    from sklearn.linear_model import LinearRegression

except:

    os.system('pip import sklearn.linear_model')

    from sklearn.linear_model import LinearRegression

try:

    from sklearn.preprocessing import MinMaxScaler

except:

    os.system('pip install sklearn.preprocessing')

    from sklearn.preprocessing import MinMaxScaler

try:

    import scipy

except:

    os.system('pip install scipy')

    import scipy

try:

    import tensorflow

except:

    os.system('pip install tensorflow')

    import tensorflow

try:

    from keras.models  import Sequential 

except:

    os.system('pip install keras')

    from keras.models  import Sequential 

from keras.layers import Dense, LSTM 




style.use('fivethirtyeight')

plt.rcParams['axes.formatter.useoffset'] = False

pd.plotting.register_matplotlib_converters()

######################################
# Functions
######################################
#------------------------------------#
def predict_prices(dates, prices, x):
#------------------------------------#
    print("tools_predict_generic: predict_prices ENTRY")
    svr_lin = SVR(kernel = 'linear', C=1e3)
    svr_rbf = SVR(kernel = 'rbf', C=1e3, gamma =  0.0004)
    svr_poly = SVR(kernel = 'poly', C = 1e3, degree = 2)

    #print("tools_predict_generic: predict_prices starting to svr_lin.fit data")
    #svr_lin.fit(dates, prices)
    
    print("tools_predict_generic: predict_prices starting to svr_poly.fit data")
    svr_poly.fit(dates, prices)
    poly_predict_feature = svr_poly.predict(x)[0]

    print("tools_predict_generic: predict_prices starting to rbf_poly.fit data")
    svr_rbf.fit(dates, prices)
    rbf_predict_feature = svr_rbf.predict(x)[0]

    plt.scatter(dates, prices, color='black', label = 'Actual Data')
    plt.scatter( dates, svr_poly.predict(dates), color = 'green', label = 'Poly Model', alpha = 0.1)
    plt.scatter( dates, svr_rbf.predict(dates), color = 'red', label = 'RBF Model')
    plt.plot(x, poly_predict_feature, '*', color = 'green', label = 'Poly Predict_' + FEATURE_VARIABLE)
    plt.plot( x, rbf_predict_feature, '*', color = 'blue', label = 'RBF Predict_' + FEATURE_VARIABLE)

    plt.xlabel('Date')
    plt.ylabel('Prices')
    plt.title(ax1_subject + " Support Vector Regression")
    plt.legend()


    print("Poly answer for predicting the "  + FEATURE_VARIABLE + " record", str(x), "is:",poly_predict_feature)
    print("RBF answer for predicting the "  + FEATURE_VARIABLE + " record", str(x), "is:", rbf_predict_feature )#svr_rbf.predict(x)[0])

    plt.show()

    return  poly_predict_feature, rbf_predict_feature




# HOUSE KEEPING START HERE
######################################

currPath = os.getcwd()              # Directory you are in NOW

savePath = 'askew'                  # We will be creating this new sub-directory

myPath = (currPath + '/' + savePath)# The full path of the new sub-dir

dir_path = os.path.dirname(os.path.realpath(__file__))


if __name__ == '__main__':

    df = pd.DataFrame()

    if len(sys.argv) > 1:
        
        if sys.argv[1]:
        
           df = sys.argv[1]
        
    else:
        
        try:

            os.chdir(myPath)

            ax1_subject = 'JCP'

            df = pd.read_csv((ax1_subject + '.csv'), parse_dates=True, index_col =0)

        except Exception as e:

            print("stocks_1.py did not find", ax1_subject, "information. As it's not in the datawarehouse, is the ticker symbol spelled correctly?")

            print(e)

            popupmsg("Symbol " + ax1_subject + " is not found! -- Spelling?")

            sys.exit(0)

#######################################
# MAIN LOGIC
#######################################
DAYS_TO_REGRESS = 60

FEATURE_VARIABLE = 'Close'

df_save_last = df.tail(1) 

#df = df.head(len(df) -1)
df = df.iloc[-(DAYS_TO_REGRESS):-1]

df.reset_index(inplace = True)

df_dates = df.loc[:, 'Date']

df_FEATURE = df.loc[:, FEATURE_VARIABLE]

dates = []

prices = []

#-------------------------------------#
# Use cnt instead of day of month, as 
#     we can read in more than 1 month
#     or we start on mid-month (from 15nth
#     thru the next 14nth.)
#-------------------------------------#

for x in range(len(df_dates)):

        dates.append([x+1])

for open_price in df_FEATURE:

    prices.append(float(open_price))


poly_predict_feature, rbf_predict_feature = predict_prices(dates, prices, [[(DAYS_TO_REGRESS + 1)]])