Assignment2 file:///home/oslab-09/Downloads/Assignment2.

html

In [6]: import pandas as pd

df=pd.read_csv('temperatures.csv')

In [8]: df.columns

Out[8]: Index(['YEAR', 'JAN', 'FEB', 'MAR', 'APR', 'MAY', 'JUN', 'JUL', 'AUG', 'SEP',
'OCT', 'NOV', 'DEC', 'ANNUAL', 'JAN-FEB', 'MAR-MAY', 'JUN-SEP',
'OCT-DEC'],
dtype='object')

In [14]: import pandas as pd

from sklearn.linear_model import LinearRegression
from sklearn.metrics import mean_squared_error, mean_absolute_error, r2_score

# Assuming df is already loaded

# df = pd.read_csv('temperature.csv')

months = ['JAN', 'FEB', 'MAR', 'APR', 'MAY', 'JUN',

'JUL', 'AUG', 'SEP', 'OCT', 'NOV', 'DEC']
# this is for the 12 months calculate the 12 different models

# Reshape YEAR column once here: convert Series to 2D numpy array (n_samples, 1)
X_year = df['YEAR'].values.reshape(-1, 1)

# store the 12 models and 12 performances

models = {}
performance = {}

# Train models
for month in months:
y = df[month]# Select temperature data for this month as target variable y
model = LinearRegression()# Create a new Linear Regression model
model.fit(X_year, y) # Train the model using X_year (year) as input and y (temperature) as output
models[month] = model# Store the trained model in a dictionary with month as key

# Evaluate models
for month in months:
y_true = df[month]
y_pred = models[month].predict(X_year)

1 of 6 04/08/25, 15:53
Assignment2 file:///home/oslab-09/Downloads/Assignment2.html

#MSE-The average of the squares of the errors — the difference between actual and predicted values.
#MAE-The average of the absolute differences between actual and predicted values.
mse = mean_squared_error(y_true, y_pred)
mae = mean_absolute_error(y_true, y_pred)
r2 = r2_score(y_true, y_pred)

performance[month] = {'MSE': mse, 'MAE': mae, 'R2': r2}

# Print performance
for month in months:
perf = performance[month]
print(f"{month} -> MSE: {perf['MSE']:.3f}, MAE: {perf['MAE']:.3f}, R2: {perf['R2']:.3f}")

JAN -> MSE: 0.462, MAE: 0.527, R2: 0.331

FEB -> MSE: 0.763, MAE: 0.679, R2: 0.419
MAR -> MSE: 0.785, MAE: 0.714, R2: 0.307
APR -> MSE: 0.555, MAE: 0.585, R2: 0.292
MAY -> MSE: 0.434, MAE: 0.510, R2: 0.166
JUN -> MSE: 0.342, MAE: 0.482, R2: 0.138
JUL -> MSE: 0.168, MAE: 0.315, R2: 0.229
AUG -> MSE: 0.129, MAE: 0.282, R2: 0.428
SEP -> MSE: 0.164, MAE: 0.312, R2: 0.441
OCT -> MSE: 0.322, MAE: 0.428, R2: 0.347
NOV -> MSE: 0.260, MAE: 0.380, R2: 0.487
DEC -> MSE: 0.282, MAE: 0.381, R2: 0.536

In [15]: import matplotlib.pyplot as plt

months_to_plot = ['JAN', 'FEB', 'OCT', 'DEC']

plt.figure(figsize=(12, 10))

for i, month in enumerate(months_to_plot, 1):

y_true = df[month]
y_pred = models[month].predict(X_year)

plt.subplot(2, 2, i) # 2 rows, 2 cols grid

plt.scatter(df['YEAR'], y_true, color='blue', label='Actual')
plt.plot(df['YEAR'], y_pred, color='red', label='Predicted')
plt.title(f'Linear Regression for {month}')
plt.xlabel('Year')

2 of 6 04/08/25, 15:53
Assignment2 file:///home/oslab-09/Downloads/Assignment2.html

plt.ylabel('Temperature (°C)')
plt.legend()

plt.tight_layout()
plt.show()

3 of 6 04/08/25, 15:53
Assignment2 file:///home/oslab-09/Downloads/Assignment2.html

4 of 6 04/08/25, 15:53
Assignment2 file:///home/oslab-09/Downloads/Assignment2.html

In [16]: from statistics import mean

mean_mse = mean(performance[m]['MSE'] for m in months)

mean_mae = mean(performance[m]['MAE'] for m in months)
mean_r2 = mean(performance[m]['R2'] for m in months)

print("Average performance across all months:")

print(f"Mean MSE: {mean_mse:.3f}")#Lower MSE → better fit.
print(f"Mean MAE: {mean_mae:.3f}")#Lower MAE → better fit.
print(f"Mean R2 : {mean_r2:.3f}")#Closer to 1 → better model.Near 0 → model does no better than average prediction.

Average performance across all months:

Mean MSE: 0.389
Mean MAE: 0.466
Mean R2 : 0.343

In [17]: def predict_temperature(year, month):

"""
Predict temperature for a given year and month using the trained model.

Parameters:
- year: int or float, the year to predict temperature for
- month: str, month abbreviation in uppercase, e.g. 'JAN', 'FEB', etc.

Returns:
- predicted temperature (float)
"""
month = month.upper()
if month not in models:
return f"Error: Month '{month}' model not found. Choose from: {list(models.keys())}"

model = models[month]
year_reshaped = [[year]] # Reshape to 2D array as sklearn expects

pred_temp = model.predict(year_reshaped)[0] # Predict and get scalar value

return pred_temp

# Example usage:
print(predict_temperature(2025, 'JAN'))

5 of 6 04/08/25, 15:53
Assignment2 file:///home/oslab-09/Downloads/Assignment2.html

print(predict_temperature(1990, 'OCT'))

24.622018013157692
30.146403088112617

In [ ]:

6 of 6 04/08/25, 15:53