Question

以下是csv格式的以下数据集的片段：

quantity revenue     time_x    transaction_id   user_id
    1       0       57:57.0 0        0             0
    1       0       18:59.0 0        1

我想在user_id为空时删除整行。我怎么在python中这样做？到目前为止，这是我的代码：

activity = pd.read_csv("activity(delimited).csv", delimiter=';', error_bad_lines=False, dtype=object)
impression = pd.read_csv("impression(delimited).csv", delimiter=';', error_bad_lines=False, dtype=object)
click = pd.read_csv("click(delimited).csv", delimiter=';', error_bad_lines=False, dtype=object)

pre_merge = activity.merge(impression, on="user_id", how="outer")
merged = pre_merge.merge(click, on="user_id", how="outer")
merged.to_csv("merged.csv", index=False)

open_merged = pd.read_csv("merged.csv", delimiter=',', error_bad_lines= False, dtype=object)

filtered_merged = open_merged.dropna(axis='columns', how='all')

另外，如何以有效的方式编写代码？

Answer 1

与熊猫：

import pandas as pd

df = pd.read_csv("path/to/csv/data.csv", delimiter=';', error_bad_lines=False)
df = df[pd.notnull(df.user_id)] # boolean indexing

# Shift user_id to first column
df = df.set_index("user_id")
df = df.reset_index()

df.to_csv("path/to/csv/data.csv", index=False)

括号表示法允许您提供可迭代的布尔值。这称为boolean indexing。在numpy，matlab和R

中使用了类似的概念和语法

Answer 2

不同风格：获取数据，加入然后删除。保持名称空间干净。

activity   = pd.read_csv("activity(delimited).csv", delimiter=';', error_bad_lines=False)
impression = pd.read_csv("impression(delimited).csv", delimiter=';', error_bad_lines=False)
pre_merge  = activity.merge(impression, on="user_id", how="outer")
del activity, impression

click      = pd.read_csv("click(delimited).csv", delimiter=';', error_bad_lines=False)
merged     = pre_merge.merge(click, on="user_id", how="outer")
merged.to_csv("merged.csv", index=False)
del click

open_merged = pd.read_csv("merged.csv", error_bad_lines= False)

filtered_merged = open_merged.dropna(axis='columns', how='all')

如何在csv文件中的列上删除整行数据集？

2 个答案: