code

Created on Feb 24 2024
@author Fadhaa Ali

import urllib3
import gc
import warnings
warnings.filterwarnings("ignore")
import pandas as pd
urllib3.disable_warnings()
import matplotlib.pyplot as plt
import numpy as np
from statsmodels.tsa.arima.model import ARIMA
from sklearn.metrics import mean_squared_error


class Pred:
    def __init__(self):
        return

    def getData(self, fi1):
        htt = urllib3.PoolManager()
        ml = htt.request('GET', fi1)
        ml = ml.data
        da = str(ml)
        return da

    def reformat(self, da):
        i = 0
        FD = []
        lis = []
        pc = []
        dt = []
        no_days_price = da.count('test="relative-most-active-last">')
        no_date = da.count('<time dateTime=')
        ix2 = 2

        for i in range(no_days_price):
            ix1 = da.index('test="relative-most-active-last">', ix2, len(da)) + 33
            ix2 = ix1 + 6
            pc.append(da[ix1:ix2])

        ix2 = 2

        for i in range(no_date):
            ix1 = da.index('<time dateTime=', ix2, len(da)) + 16
            ix2 = ix1 + 12
            dt.append(da[ix1:ix2])

        k = 0

        for i in range(1, no_date):
            FD.append([dt[i], pc[k], pc[k+1], pc[k+2]])
            k += 3

        ##FD.reverse()
        Reformated_data = {
            "Date": [FD[i][0] for i in range(len(FD))],
            "Open": [float(FD[i][1]) for i in range(len(FD))],
            "High": [float(FD[i][2]) for i in range(len(FD))],
            "Low": [float(FD[i][3]) for i in range(len(FD))]
        }

        PriceData = pd.DataFrame(Reformated_data)
        return PriceData

    def estim(self, PriceData):
        model = ARIMA(PriceData['Low'], order=(1, 1, 1))
        model_fit = model.fit()
        forecast = model_fit.get_forecast(steps=30)

        train_size = int(len(PriceData) * 0.8)
        train, test = PriceData[0:train_size], PriceData[train_size:len(PriceData)]

        model_train = ARIMA(train['Low'], order=(1, 1, 1))
        model_train_fit = model_train.fit()
        test_forecast = model_train_fit.get_forecast(steps=len(test))
        test_forecast_series = pd.Series(test_forecast.predicted_mean, index=test.index)

        mse = mean_squared_error(test['Low'], test_forecast_series)
        rmse = mse ** 0.5

        model2 = ARIMA(PriceData['High'], order=(1, 1, 1))
        model_fit2 = model2.fit()
        forecast2 = model_fit2.get_forecast(steps=30)

        train_size2 = int(len(PriceData) * 0.8)
        train2, test2 = PriceData[0:train_size2], PriceData[train_size2:len(PriceData)]

        model_train2 = ARIMA(train2['High'], order=(1, 1, 1))
        model_train_fit2 = model_train2.fit()
        test_forecast2 = model_train_fit2.get_forecast(steps=len(test2))
        test_forecast_series2 = pd.Series(test_forecast2.predicted_mean, index=test2.index)

        mse2 = mean_squared_error(test2['High'], test_forecast_series2)
        rmse2 = mse2 ** 0.5

        plt.figure(figsize=(14, 7))
        plt.plot(train['Low'], label='Training Low', color='red')
        plt.plot(test['Low'], label='Actual Low', color='red')
        plt.plot(test_forecast_series, label='Forecasted Low', color='yellow')
        plt.plot(train2['High'], label='Training High', color='blue')
        plt.plot(test2['High'], label='Actual High', color='blue')
        plt.plot(test_forecast_series2, label='Forecasted High', color='green')
        plt.fill_between(test.index, test_forecast.conf_int().iloc[:, 0],
                         test_forecast.conf_int().iloc[:, 1], color='k', alpha=.15)
        plt.fill_between(test2.index, test_forecast2.conf_int().iloc[:, 0],
                         test_forecast2.conf_int().iloc[:, 1], color='k', alpha=.15)
        plt.title('ARIMA Model Evaluation')
        plt.xlabel('Day')
        plt.ylabel('Price')
        plt.legend()
        plt.show()

        return [rmse, rmse2]