pandas

根据其官方网站介绍，pandas（是的，小写“p”）是“一种基于 Python 编程语言构建的快速、强大、灵活且易于使用的开源数据分析与操作工具”。对于熟悉 Excel 的用户来说，pandas 是在 Python 中处理数据最直观的工具，因为 pandas 的 DataFrame 与 Excel 表格有许多相似之处。

1import pandas as pd
2
3a = pd.Series([10, 20, 30, 40])
4print(a)

1import pandas as pd
2
3df = pd.read_csv('microsoft_stock.csv')
4
5print(df)

![三个 DataFrame 操作]

1import pandas as pd
2
3df = pd.read_csv('microsoft_stock.csv')
4df = df[df['Close'] >= 100]
5
6print(df)

1import pandas as pd
2
3df = pd.read_csv('microsoft_stock.csv')
4
5df = df[(df['Open'] >= 100) & (df['Close'] < 200)]
6# 行数据需同时满足两个条件
7
8print(df)

1import pandas as pd
2
3df = pd.read_csv('microsoft_stock.csv')
4
5df = df[(df['Open'] >= 100) | (df['Close'] < 200)]
6# 行数据只需满足任一条件
7
8print(df)

1import pandas as pd
2import datetime
3
4df = pd.read_csv('microsoft_stock.csv')
5
6# 修改数据类型
7df["Date"] = pd.to_datetime(df['Date'], format='%m/%d/%Y %H:%M:%S')
8
9# 第一种情况：Date 仅为普通列，非索引
10df1 = df[df['Date'] > datetime.date(2016, 1, 1)]
11print(df1)
12
13# 第二种情况：将 Date 设为 DataFrame 的索引
14df = df.set_index('Date')
15df2 = df.loc['2016-01-01':]
16print(df2)

1import pandas as pd
2
3df = pd.read_csv('microsoft_stock.csv')
4
5# 修改数据类型
6df["Date"] = pd.to_datetime(df['Date'], format='%m/%d/%Y %H:%M:%S')
7 
8# 聚合操作
9df_agg = df.groupby("Open")["Close"].max()
10
11print(df_agg)

1import pandas as pd
2
3df = pd.read_csv('microsoft_stock.csv')
4
5# 修改数据类型
6df["Date"] = pd.to_datetime(df['Date'], format='%m/%d/%Y %H:%M:%S')
7 
8# 按周进行聚合
9df_agg = df.groupby(pd.Grouper(key='Date', freq='W')).max()
10
11print(df_agg)

1import pandas as pd
2
3df = pd.read_csv('microsoft_stock.csv')
4
5# 修改数据类型
6df["Date"] = pd.to_datetime(df['Date'], format='%m/%d/%Y %H:%M:%S')
7 
8# 多重聚合操作
9df_agg = df.groupby(pd.Grouper(key='Date', freq='W')).agg(
10        {
11        'Open': 'max',
12        'High': 'max',
13        'Low': 'min',
14        'Close': 'max',
15        'Volume': 'sum'
16        }
17    )
18
19print(df_agg)

1import pandas as pd
2
3stocks = pd.read_csv('microsoft_stock.csv')
4weather = pd.read_csv('seattle_weather.csv')
5
6# 修改数据类型
7stocks["Date"] = pd.to_datetime(stocks['Date'], format='%m/%d/%Y %H:%M:%S').dt.date
8weather["date"] = pd.to_datetime(weather['date'], format='%Y-%m-%d').dt.date
9
10# 数据合并
11df = stocks.merge(weather, how='left', left_on='Date', right_on='date')
12
13print(weather)
14print(stocks)
15print(df)