# Let us get started with some basic setup
import warnings; warnings.filterwarnings("ignore")
import pandas as pd
import matplotlib.pyplot as plt
 
 

# Data setup
url = "https://raw.githubusercontent.com/jbrownlee/Datasets/master/daily-min-temperatures.csv"
df = pd.read_csv(url)
print("Shape of DataFrame:", df.shape)
print(df.head())
 
 

# Let us plot the temperature data
plt.figure(figsize=(10,4))
plt.plot(pd.to_datetime(df["Date"]), df["Temp"])
plt.title("Daily Minimum Temperatures")
plt.xlabel("Date")
plt.ylabel("Temperature (Celsius)")
plt.show()

# Let us convert the date column and set as index for easy analysis
df["Date"] = pd.to_datetime(df["Date"])
df = df.set_index("Date")
 
 

# Let us see if our index is a DatetimeIndex now
print(type(df.index))
print(df.index[:5])

# Decompose with statsmodels
from statsmodels.tsa.seasonal import seasonal_decompose
result = seasonal_decompose(df["Temp"], model="additive", period=365)
 
 

# Let us plot all the parts of decomposition
result.plot()
plt.tight_layout()
plt.show()

# Let us access just the trend as an example
trend = result.trend
plt.plot(trend, color="red")
plt.title("Trend Component Only")
plt.ylabel("Temperature")
plt.show()

# Plotting the seasonal component alone
seasonal = result.seasonal
plt.plot(seasonal, color="green")
plt.title("Seasonal Component Only")
plt.show()

# Try plotting only the residual for a closer look at randomness
residual = result.resid
residual.plot(style=".", alpha=0.7)
plt.title("Residual (Random) Component Only")
plt.show()

# Safe Access: Checking for missing values in each part
print("Missing in trend:", trend.isna().sum())
print("Missing in seasonal:", seasonal.isna().sum())
print("Missing in residual:", residual.isna().sum())

# Fill missing values if needed, using 'bfill' (backward fill) method
trend_filled = trend.fillna(method="bfill")
seasonal_filled = seasonal.fillna(method="bfill")
residual_filled = residual.fillna(method="bfill")

# Using Decomposition in a Mini Project: Finding the hottest months
monthly = df.resample("M").mean()
seasonal_month = seasonal.groupby(seasonal.index.month).mean()
plt.bar(range(1,13), seasonal_month)
plt.title("Average Seasonal Effect by Month")
plt.xlabel("Month")
plt.ylabel("Seasonal Offset")
plt.show()

# Try your own! Ask the user for a month and report the seasonal effect
month = input("Enter a month number (1-12): ")
try:
    num = int(month)
    if 1 <= num <= 12:
        print(f"Seasonal effect for month {num}: {seasonal_month[num]}")
    else:
        print("Please pick a number from 1 to 12.")
except ValueError:
    print("That was not a valid number.")
    

# Best Practices: Try using multiplicative mode if your data multiplies, not adds
result_mul = seasonal_decompose(df["Temp"], model="multiplicative", period=365)
result_mul.plot()
plt.suptitle("Multiplicative Decomposition", y=1.02)
plt.show()

# Troubleshooting: What if your period is wrong?
try:
    seasonal_decompose(df["Temp"], model="additive", period=7)
    print("No error, but the patterns might look strange!")
except Exception as e:
    print("Error:", e)
    

# Extra Tip: Save your trend to a new file
trend_filled.to_csv("trend_component.csv")
print("Trend component saved!")