python - 从具有非数字索引的数据框中删除行

Question

我一直在使用 pandas 对 CSV 文件进行一些有趣的过滤，但遇到了障碍。我正在尝试检查索引列中是否存在乱码文本(非整数)数据，并删除这些行。我尝试在导入时使用条件将它们从数据框中删除，并且之后我尝试迭代它们但没有成功。这是一个例子:

df = pd.read_csv(file, encoding='cp1252').set_index("numbers")
results = df[df["columnA"].str.contains("search_data") & ~df["columnB"].isin(seach_list)]
#I need to add to the above statement to check column "numbers" which I have set to be the index,
#to catch some expected garbled text and filter it out... because it is
#an integer, I can't use str.contains or isdigit or isalnum, I've tried to do len(df["columns"] < 20 , df.index < 20 .... i've tried 
#i've tried a few other options at this point as well
# after bringing it in, I've also tried iterating through it:
#
for index, row in results.iterrows():
    if not (isinstance( row["numbers"], int )):
         print(str(row["numbers"]))
         #append whole row to new dataframe
#This also didn't work   

对我能做什么有什么想法吗？

Example data in the "numbers columns = 329381432
Example garbled text in "numbers" column that I am 
trying to keep from importing: äu$ÒÔ”5$ò"Â$”äu$ÒÔ”5$ò 

作为旁注，我必须更改 pd 函数的编码，以便当存在一些非 utf-8 数据时我仍然可以读取文件中的所有好数据...否则会抛出错误导入。

Answer 1

您可以使用pd.to_numeric将numbers列转换为数字。所有非数字条目都将被强制为 NaN，然后您可以删除这些行。

df = pd.read_csv(file, encoding='cp1252')
df['numbers'] = pd.to_numeric(df['numbers'], errors='coerce')

df = df.dropna(subset=['numbers']).set_index('numbers')