import warnings
warnings.filterwarnings("ignore")

# Suppress warnings for a smooth learning experience

# Data setup: Let us get a real-world time series!
import pandas as pd

url = "https://raw.githubusercontent.com/jbrownlee/Datasets/master/airline-passengers.csv"
data = pd.read_csv(url, parse_dates=['Month'])
print("Shape:", data.shape)
print("First 5 rows:")
print(data.head())

# Plot the time series to see the pattern
import matplotlib.pyplot as plt

plt.figure(figsize=(10, 4))
plt.plot(data['Month'], data['Passengers'], marker='o')
plt.title("Monthly Airline Passengers")
plt.xlabel("Month")
plt.ylabel("Passengers")
plt.grid(True)
plt.tight_layout()
plt.show()

# Let us split the data into training and testing sets
from sklearn.model_selection import train_test_split

values = data['Passengers'].values
train_size = int(len(values) * 0.7)
train, test = values[:train_size], values[train_size:]

print(f"Training points: {len(train)} | Testing points: {len(test)}")

# Create a simple prediction: Use the last seen value (naive forecast)
predictions = []
history = list(train)

for t in range(len(test)):
    yhat = history[-1]
    predictions.append(yhat)
    history.append(test[t])

print(f"First 5 predictions: {predictions[:5]}")

# Let us see how accurate our predictions are
from sklearn.metrics import mean_squared_error
import numpy as np

mse = mean_squared_error(test, predictions)
rmse = np.sqrt(mse)

print(f"RMSE: {rmse:.2f}")

# Visualize: Compare predictions to real values
plt.figure(figsize=(10,4))
plt.plot(data['Month'][-len(test):], test, label="Actual")
plt.plot(data['Month'][-len(test):], predictions, label="Predicted", linestyle="--")
plt.title("Walk-Forward Predictions vs. Actuals")
plt.xlabel("Month")
plt.ylabel("Passengers")
plt.legend()
plt.grid(True)
plt.tight_layout()
plt.show()

# Walk-forward with a simple moving average (window=3)
window = 3
pred_ma = []
history_ma = list(train)

for t in range(len(test)):
    if len(history_ma) < window:
        yhat_ma = history_ma[-1]
    else:
        yhat_ma = np.mean(history_ma[-window:])
    pred_ma.append(yhat_ma)
    history_ma.append(test[t])

print("First 5 moving average predictions:", pred_ma[:5])

# Check the accuracy of the moving average forecast
mse_ma = mean_squared_error(test, pred_ma)
rmse_ma = np.sqrt(mse_ma)
print(f"RMSE (Moving Average): {rmse_ma:.2f}")

# Plot moving average predictions vs. real values
plt.figure(figsize=(10,4))
plt.plot(data['Month'][-len(test):], test, label="Actual")
plt.plot(data['Month'][-len(test):], pred_ma, label="Moving Average Prediction", linestyle="--")
plt.title("Moving Average Walk-Forward Predictions")
plt.xlabel("Month")
plt.ylabel("Passengers")
plt.legend()
plt.grid(True)
plt.tight_layout()
plt.show()

# Using a smart model: ARIMA (quick version for demonstration)
from statsmodels.tsa.arima.model import ARIMA

history_arima = list(train)
pred_arima = []

for t in range(len(test)):
    model = ARIMA(history_arima, order=(1,1,0))
    model_fit = model.fit()
    yhat_arima = model_fit.forecast()[0]
    pred_arima.append(yhat_arima)
    history_arima.append(test[t])

print("First 5 ARIMA predictions:", pred_arima[:5])

# Check ARIMA prediction error
mse_arima = mean_squared_error(test, pred_arima)
rmse_arima = np.sqrt(mse_arima)
print(f"RMSE (ARIMA): {rmse_arima:.2f}")

# Try it yourself! Walk-forward on a different dataset
url2 = "https://raw.githubusercontent.com/jbrownlee/Datasets/master/shampoo.csv"
data2 = pd.read_csv(url2)
print("Shampoo sales preview:\n", data2.head())

values2 = data2['Sales'].values
size2 = int(len(values2) * 0.7)
train2, test2 = values2[:size2], values2[size2:]

preds2 = []
hist2 = list(train2)
for t in range(len(test2)):
    yhat2 = np.mean(hist2[-3:]) if len(hist2) >= 3 else hist2[-1]
    preds2.append(yhat2)
    hist2.append(test2[t])

rmse2 = np.sqrt(mean_squared_error(test2, preds2))
print(f"RMSE: {rmse2:.2f}")

# Challenge: Let the user pick a forecast method
print("Choose method: 1 for last value, 2 for moving average")
choice = input("Enter 1 or 2: ")
if choice == "1":
    print("You picked: Last value (naive)")
    preds_challenge = []
    hist_challenge = list(train)
    for t in range(len(test)):
        yhat = hist_challenge[-1]
        preds_challenge.append(yhat)
        hist_challenge.append(test[t])
elif choice == "2":
    print("You picked: Moving average (window=3)")
    preds_challenge = []
    hist_challenge = list(train)
    for t in range(len(test)):
        if len(hist_challenge) < 3:
            yhat = hist_challenge[-1]
        else:
            yhat = np.mean(hist_challenge[-3:])
        preds_challenge.append(yhat)
        hist_challenge.append(test[t])
else:
    print("Please enter only 1 or 2.")

if choice in ["1", "2"]:
    error = np.sqrt(mean_squared_error(test, preds_challenge))
    print(f"Your selected method RMSE: {error:.2f}")