# Suppress warnings for a cleaner notebook.
import warnings; warnings.filterwarnings("ignore")
 
# Import libraries for data and plotting.
import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
np.random.seed(42)
%matplotlib inline

# Data setup: load airline passengers monthly dataset.
url = "https://raw.githubusercontent.com/jbrownlee/Datasets/master/airline-passengers.csv"
df = pd.read_csv(url)

print("Data shape:", df.shape)
df.head()

# First look: line plot of passengers over time.
plt.figure(figsize=(10,4))
plt.plot(df['Month'], df['Passengers'], marker='o')
plt.title("Monthly Airline Passengers 1949-1960")
plt.xlabel("Month")
plt.ylabel("Passengers")
plt.xticks(rotation=45)
plt.tight_layout()
plt.show()

# Probability in action: random coin flip simulation.
np.random.seed(42)
flips = np.random.choice(["Heads", "Tails"], size=10, p=[0.5, 0.5])
print("Coin flips:", flips)

# Descriptive statistics: look at passenger summary.
print("How many months?", len(df))
print("Average passengers:", np.mean(df['Passengers']))
print("Smallest month:", np.min(df['Passengers']))
print("Largest month:", np.max(df['Passengers']))

# Handle missing values (if any): is the data clean?
print("Any missing data?", df.isnull().values.any())

# Fill or drop missing data if needed.
df_clean = df.dropna()

# Let us try a Bernoulli distribution simulation.
from scipy.stats import bernoulli
outcomes = bernoulli.rvs(p=0.7, size=12, random_state=42)
print("Bernoulli outcomes:", outcomes)

# Binomial distribution: number of successes in many tries.
from scipy.stats import binom
trials = 15
prob_success = 0.6
binom_samples = binom.rvs(trials, prob_success, size=8, random_state=42)
print("Binomial samples:", binom_samples)

# Continuous distribution example: normally distributed data.
normal_data = np.random.normal(loc=0, scale=1, size=1000)
plt.hist(normal_data, bins=30, alpha=0.7)
plt.title("Histogram of 1000 random values (Normal Distribution)")
plt.xlabel("Value")
plt.ylabel("Frequency")
plt.show()

# Sampling and the Central Limit Theorem (CLT) demo.
pop = np.random.exponential(scale=1.5, size=5000)
means = [np.mean(np.random.choice(pop, 30)) for _ in range(200)]
plt.hist(means, bins=25, alpha=0.7)
plt.title("Sample Means from Exponential Population")
plt.xlabel("Sample mean")
plt.ylabel("Count")
plt.show()

# Confidence interval for mean passengers.
import scipy.stats as stats
mu = np.mean(df['Passengers'])
sigma = np.std(df['Passengers'], ddof=1)
n = len(df['Passengers'])
conf_int = stats.t.interval(0.95, n-1, loc=mu, scale=sigma/np.sqrt(n))
print("95% confidence interval for mean passengers:", conf_int)

# Hypothesis test: Did passengers grow from 1949 to 1960?
early = df[df['Month'] < '1955']['Passengers']
late = df[df['Month'] >= '1955']['Passengers']
t_stat, p_value = stats.ttest_ind(early, late, equal_var=False)
print("P-value:", p_value)

# Correlation and covariance: link between time and passengers.
df["MonthNum"] = np.arange(len(df))
corr = np.corrcoef(df["MonthNum"], df["Passengers"])[0,1]
cov = np.cov(df["MonthNum"], df["Passengers"])[0,1]
print("Correlation:", corr)
print("Covariance:", cov)

# Simple linear regression: predict next month's passengers.
from sklearn.linear_model import LinearRegression
model = LinearRegression()
X = df[["MonthNum"]]
y = df["Passengers"]
model.fit(X, y)
preds = model.predict(X)
plt.plot(df["MonthNum"], y, label="Actual")
plt.plot(df["MonthNum"], preds, label="Predicted")
plt.legend()
plt.title("Linear Regression Fit")
plt.xlabel("Month Number")
plt.ylabel("Passengers")
plt.show()

# Logistic regression preview: Titanic dataset survival prediction.
import seaborn as sns
titanic = sns.load_dataset("titanic")
titanic = titanic.dropna(subset=["age", "fare"])
from sklearn.linear_model import LogisticRegression
X_titanic = titanic[["age", "fare"]]
y_titanic = titanic["survived"]
logr = LogisticRegression(max_iter=200)
logr.fit(X_titanic, y_titanic)
score = logr.score(X_titanic, y_titanic)
print("Accuracy:", round(score, 2))

# Time series train-test split: predict the last year.
train = df[df['Month'] < '1959-01']
test = df[df['Month'] >= '1959-01']
print("Train months:", train.shape[0])
print("Test months:", test.shape[0])

# Simple moving average forecast for test months.
test = test.copy()
window = 12
train_means = train['Passengers'].rolling(window=window).mean().dropna()
last_mean = train_means.iloc[-1]
test['SMA_Pred'] = last_mean
plt.plot(train['Month'], train['Passengers'], label='Train')
plt.plot(test['Month'], test['Passengers'], label='Test')
plt.plot(test['Month'], test['SMA_Pred'], label='SMA Prediction')
plt.legend()
plt.title('Moving Average Forecast Example')
plt.xlabel('Month')
plt.ylabel('Passengers')
plt.show()

# Bonus: Let us try a real forecasting model (ARIMA).
from statsmodels.tsa.arima.model import ARIMA
model_arima = ARIMA(train['Passengers'], order=(1,1,1))
fit_arima = model_arima.fit()
forecast = fit_arima.forecast(steps=len(test))
test = test.copy()
test['ARIMA_Pred'] = forecast.values
plt.plot(train['Month'], train['Passengers'], label='Train')
plt.plot(test['Month'], test['Passengers'], label='Test')
plt.plot(test['Month'], test['ARIMA_Pred'], label='ARIMA Forecast')
plt.legend()
plt.title('ARIMA Model Forecast')
plt.xlabel('Month')
plt.ylabel('Passengers')
plt.show()

# Mini-project part 1: Simulate a seasonal sales pattern.
np.random.seed(100)
months = np.arange(60)
season = 20 * np.sin(2*np.pi*months/12)
trend = 3 * months
noise = np.random.normal(0, 10, 60)
sales = 100 + trend + season + noise
plt.plot(months, sales)
plt.title("Simulated Monthly Sales (5 Years)")
plt.xlabel("Month")
plt.ylabel("Sales")
plt.show()

# Mini-project part 2: Analyze daily temperature real data.
url_temp = "https://raw.githubusercontent.com/jbrownlee/Datasets/master/daily-min-temperatures.csv"
temps = pd.read_csv(url_temp)
temps['Date'] = pd.to_datetime(temps['Date'])
temps.set_index('Date', inplace=True)
temps['Temp'].plot(figsize=(10,4))
plt.title('Daily Minimum Temperatures, Melbourne')
plt.xlabel('Date')
plt.ylabel('Min Temp (C)')
plt.show()

# Extra tip: making features from dates.
temps['Month'] = temps.index.month
monthly_avg = temps.groupby('Month')['Temp'].mean()
print(monthly_avg)

# Challenge exercise: Shift temperature data and compare.
temps['Yesterday'] = temps['Temp'].shift(1)
plt.scatter(temps['Yesterday'], temps['Temp'], alpha=0.5)
plt.title('Current Temp vs Previous Day')
plt.xlabel('Yesterday Temp (C)')
plt.ylabel('Today Temp (C)')
plt.show()