Quiz tổng hợp

1# 1. Tổng revenue theo product
2df.groupby('product')['revenue'].sum()
3# A: 370, B: 670, C: 480
4
5# 2. Product có tổng revenue cao nhất
6df.groupby('product')['revenue'].sum().idxmax()
7# 'B' (670)
8
9# 3. Trung bình quantity theo region
10df.groupby('region')['quantity'].mean()
11# North: 17.0, South: 21.0

1# 1. Đếm missing values
2df.isnull().sum()
3# name: 0, age: 2, salary: 2, department: 0
4
5# 2. Điền age bằng median
6df['age'].fillna(df['age'].median(), inplace=True)
7# median(25, 30, 28) = 28 → age = [25, 28, 30, 28, 28]
8
9# 3. Điền salary bằng mean theo department
10df['salary'] = df.groupby('department')['salary'].transform(
11    lambda x: x.fillna(x.mean())
12)
13# IT mean = 5000, HR mean = 6000→7000 mean = 6500
14# salary = [5000, 6000, 5000, 7000, 5000]

1import matplotlib.pyplot as plt
2
3fig, axes = plt.subplots(1, 2, figsize=(12, 5))
4
5# Subplot 1: Bar chart revenue theo product
6revenue_by_product = df.groupby('product')['revenue'].sum()
7axes[0].bar(revenue_by_product.index, revenue_by_product.values, color=['#3498db', '#e74c3c', '#2ecc71'])
8axes[0].set_title('Tổng Revenue theo Product')
9axes[0].set_xlabel('Product')
10axes[0].set_ylabel('Revenue')
11
12# Subplot 2: Pie chart quantity theo region
13qty_by_region = df.groupby('region')['quantity'].sum()
14axes[1].pie(qty_by_region.values, labels=qty_by_region.index, autopct='%1.1f%%', colors=['#9b59b6', '#f39c12'])
15axes[1].set_title('Tỷ lệ Quantity theo Region')
16
17plt.tight_layout()
18plt.show()