数据探索

1import pandas as pd
2import numpy as np
3
4import matplotlib.pyplot as plt
5import seaborn as sns
6
7df_transactions = pd.read_csv('mock_transactions.csv', header=0)
8
9print('数据集维度')
10print(df_transactions.shape)
11
12print('\n数据集样本')
13print(df_transactions.head())  
14# df_transactions.tail() # 显示最后5条记录

1# 描述数值型列
2print(df_transactions.describe())

1# 单列操作
2col = df_transactions['InvoiceNo']
3
4print('数据类型:', type(col))

1# 每位客户的记录条数统计
2col = df_transactions['CustomerID']
3print(col.value_counts())

1plt.figure(figsize=(14, 6))
2
3sns.countplot(y=df_transactions['CustomerID'])
4
5plt.xlabel('数量')
6plt.ylabel('CustomerID')
7plt.savefig('output/output.png')

1result = df_transactions.query('Quantity >= 20')
2print(result)

1mask = df_transactions['Quantity'] >= 20
2print('步骤1：创建掩码')
3print(mask)
4
5result = df_transactions[mask]
6print('\n步骤2：使用掩码过滤')
7print(result)