import pandas as pd
def basic_pd():
# pandas get data of excel
path = "D:/test/test.csv"
# Reading csv is similar to reading excel
df = pd.DataFrame(pd.read_csv(path, header=0))
# 如果含有中文一定要设置编码方式
# df = pd.DataFrame(pd.read_csv(path, header=0, encoding='gbk'))
# df = pd.DataFrame(pd.read_excel(path, header=0))
# Show the information of excel
print("Excel的文件信息:")
print(df.info())
print("提取Excel中的第一行数据:")
# 返回第一行的一维列表
for key in df.keys():
print(key)
# 返回除去第一行的二维列表
print("提取Excel中的第一行数据:")
for value in df.values:
print(value)
count_row = df.count(axis=1)
print("统计Excel中的每一行的字段数量:")
print(count_row)
print("统计Excel中的每一列的字段数量:")
count_column = df.count(axis=0)
print(count_column)
print("提取统计字段的key:")
for key in count_column.keys():
print(key)
print("提取统计字段的values:")
for value in count_column.values:
print(value)
def table_pd():
path = "D:/test/test.csv"
path1 = "D:/test/test1.csv"
df_table = pd.DataFrame(pd.read_csv(path, header=0))
df_table1 = pd.DataFrame(pd.read_csv(path1, header=0, encoding='gbk'))
# 连接的过程是以左右表中左侧第一列值为id
print("内连接:提取左右表id相同的数据")
df_inner = pd.merge(df_table, df_table1, how='inner')
print(df_inner)
print("左连接:提取左表中全部数据,右表数据补充左表中id相同的字段和数据")
df_left = pd.merge(df_table, df_table1, how='left')
print(df_left)
print("右连接:提取右表中全部数据,左表数据补充右表中id相同的字段和数据")
df_right = pd.merge(df_table, df_table1, how='right')
print(df_right)
print("左右表相互补充字段,合并id相同的数据")
df_outer = pd.merge(df_table, df_table1, how='outer')
print(df_outer)
if __name__ == '__main__':
# pandas的基本使用
basic_pd()
# pandas的表的合并
table_pd()