处理 CSV 数据文件

记录python 使用pandas 处理 csv 文件常规程序：

1. 读取 csv 文件，获取数据：

• 1. pandas

import pandas as pd
import csv
data=pd.read_csv('zi_202105071017.csv',encoding='utf-8')
print(data.head())# 查看前5行数据
print(data.tail())# 查看最后5行数据，括号里可以指定查看行数
>>
      zi_id  fanti  struct zi_str     freq
0      1    NaN     NaN      一  1338743
1      2    NaN     NaN      丁    11857
2      3    NaN     NaN      七    14477
3      4    NaN     NaN      万    28095
4      5    NaN     NaN      丈    15697
      zi_id  fanti  struct zi_str  freq
4961   4962    NaN     NaN      龋    27
4962   4963    NaN     NaN      龙  7012
4963   4964    NaN     NaN      龚    57
4964   4965    NaN     NaN      龛   379
4965   4966    NaN     NaN      龟   647
[4966 rows x 5 columns]
==================================================================================================================
print(data['zi_str'])#访问指定的列
print(data['zi_str'].values)#取出对应的值，依次放入 list 
>>
0       一
1       丁
2       七
3       万
4       丈
Name: zi_str, Length: 4966, dtype: object
['一' '丁' '七' ... '龚' '龛' '龟']
======================================================================================================
data.drop('freq', axis=1, inplace=True)
data.drop('struct', axis=1, inplace=True)#删除指定的列
>>
      zi_id  fanti zi_str
0         1    NaN      一
1         2    NaN      丁
2         3    NaN      七
3         4    NaN      万
4         5    NaN      丈
[4966 rows x 3 columns]

• 2. csv.reader()

with open("zi_202105071017.csv",'r',encoding='utf-8') as f: 
        rows = [row for row in csv.reader(f)]
print(rows[:5])
>>
[['zi_id', 'fanti', 'struct', 'zi_str', 'freq'], ['1', '', '', '一', '1338743'], ['2', '', '', '丁', '11857'], ['3', '', '', '七', '14477'], ['4', '', '', '万', '28095']]

读取 CSV 文件并取出指定行写入新的 CSV文件

import csv
count=0
with open("out.csv", 'r',encoding='utf-8', newline='') as file:
    with open("train.csv", 'w',encoding='utf-8', newline='') as trian:
        with open("valid.csv", 'w',encoding='utf-8', newline='') as valid:
            csvreader = csv.reader(file)
            trian_csvwriter = csv.writer(trian)
            valid_csvwriter = csv.writer(valid)
            valid_csvwriter.writerow(['text','ner_tags'])
            for row in csvreader:
                if count<10001:
                    trian_csvwriter.writerow(row)
                elif count>=10001 and count<=13000:
                    valid_csvwriter.writerow(row)
                else:
                    break
                count+=1

3. 数据写入 csv:

pd.DataFrame(data=pred_label, index=range(len(pred_label))).to_csv('pred.csv')

eg2:

with open('test.csv','w',encoding="utf-8",errors='ignore',newline='')as f:
  csv_write=csv.writer(f,dialect='excel')
  csv_write.writerow(['text','ner_tags'])
  stu1=['今天天气还行','B,I,O,I,I,O']
  csv_write.writerow(stu1)