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

df = pd.read_csv("/content/precio-utilidad.txt", sep=",") df.head()

# Convertimos Consolidado como un indice en el mismo data frame df.set_index("Consolidado", inplace=True) # Convertimos objet index to datatime index df.index = pd.to_datetime(df.index) # Mostramos la ultimas 10 registros df.head()

plt.figure(figsize=(14, 8)) plt.title("Multiplo P/U Walmex") plt.ylabel("Precio/utilidad") plt.plot(df);

plt.figure(figsize=(14, 8)) train = df[df.index < pd.to_datetime("2020-11-01", format='%Y-%m-%d')] test = df[df.index > pd.to_datetime("2020-11-01", format='%Y-%m-%d')] plt.plot(train, color = "black") plt.plot(test, color = "red") plt.ylabel('Precio/utilidad') plt.xlabel('Date') plt.xticks(rotation=45) plt.title("Train/Test split for BTC Data") plt.show()

from statsmodels.tsa.statespace.sarimax import SARIMAX

y = train[' Precio-utilidad']

ARMAmodel = SARIMAX(y, order = (1, 0, 1))

ARMAmodel = ARMAmodel.fit()

y_pred = ARMAmodel.get_forecast(len(test.index)) y_pred_df = y_pred.conf_int(alpha = 0.05) y_pred_df["Predictions"] = ARMAmodel.predict(start = y_pred_df.index[0], end = y_pred_df.index[-1]) y_pred_df.index = test.index y_pred_out = y_pred_df["Predictions"]

plt.plot(train, color = "black") plt.plot(test, color = "red") plt.plot(y_pred_out, color='green', label = 'Predictions') plt.ylabel('Precio/utilidad') plt.xlabel('Date') plt.xticks(rotation=45) plt.title("Train/Test split for BTC Data") plt.legend()

import numpy as np from sklearn.metrics import mean_squared_error arma_rmse = np.sqrt(mean_squared_error(test[" Precio-utilidad"].values, y_pred_df["Predictions"])) print("RMSE: ",arma_rmse)

from statsmodels.tsa.arima.model import ARIMA

ARIMAmodel = ARIMA(y, order = (2, 2, 2)) ARIMAmodel = ARIMAmodel.fit() y_pred = ARIMAmodel.get_forecast(len(test.index)) y_pred_df = y_pred.conf_int(alpha = 0.05) y_pred_df["Predictions"] = ARIMAmodel.predict(start = y_pred_df.index[0], end = y_pred_df.index[-1]) y_pred_df.index = test.index y_pred_out = y_pred_df["Predictions"] plt.plot(train, color = "black") plt.plot(test, color = "red") plt.plot(y_pred_out, color='Yellow', label = 'ARIMA Predictions') plt.legend() import numpy as np from sklearn.metrics import mean_squared_error arma_rmse = np.sqrt(mean_squared_error(test[" Precio-utilidad"].values, y_pred_df["Predictions"])) print("RMSE: ",arma_rmse)

# Metricas from sklearn.metrics import mean_absolute_error def mean_absolute_percentage_error(y_true, y_predict): return np.mean(np.abs(y_true - y_predict / y_true)) * 100 # Media movil def moving_average(series, n): """Calcula la media móvil de una serie de tiempo Arguments: ----- series -- np.array con los datos de la serie n -- el retraso (lag) de la ventana para calcular la media movil Returns ------: Un escalar con la media para la ventana indicada. """ return np.average(series[-n:]) moving_average(df, 24)

def plot_MA(series, window, plot_intervals=False, scale=1.96, plot_anormalies=False): """ plot_intervals - para graficar el error plot_anormalies - nos ayuda a hacer zoom, ver la anomlias """ rolling_mean = series.rolling(window=window).mean() # pandas plt.figure(figsize=(15,5), ) plt.plot(rolling_mean, 'g', label='Tendencia') if plot_intervals: mae = mean_absolute_error(series[window:], rolling_mean[window:]) dev = np.std(series[window:]-rolling_mean[window:]) upper_ =rolling_mean + (mae + scale*dev) lower_ =rolling_mean - (mae + scale*dev) plt.plot(upper_, 'r--', label='Límites superior/inferior') plt.plot(lower_, 'r--') if plot_anormalies: anomalies = pd.DataFrame(index=series.index, columns=series.columns) anomalies[series < lower_] = series[series < lower_] anomalies[series > upper_] = series[series > upper_] plt.plot(anomalies, 'ro', markersize=10) plt.plot(series[window:], label='Valores reales') plt.legend(loc='upper left') plt.grid(True)

plot_MA(df, 4, plot_intervals=True)