我正在尝试通过两个字符串名称过滤数据框,但问题是字符串可以在任何一个数据框的系列中,并且系列的数量是可变的。如何过滤数据帧的每个系列,然后将它们合并为一个数据框?
import pandas as pd
import os
# Directories of Statements:
cdir = "Current Directory"
odir = "Output Directory"
# Find all CSVs in cdir:
excels = [filename for filename in os.listdir(cdir) if filename.endswith(".csv")]
# Define concat_csv Function:
def concat_csv(csv_file):
df_csv = pd.read_csv(os.path.join(cdir, csv_file), header=None, index_col=None) # Load CSV into dataframe
df_final = pd.DataFrame() # Create empty dataframe
for col in df_csv: # For all columns in the dataframe filter rows by string 1 or 2 then create new dataframe
df_i = df_csv[(df_csv[col].str.contains("string1")==True) or (df_csv[col].str.contains("string2")==True)] # Use row if string equals string 1 or 2
df_final = df_final.concat(df_i, axis=1) # Concat all rows that contain string 1 or 2 to a new dataframe
# Send final dataframe to CSV in output directory:
df_final.to_csv(os.path.join(odir, os.path.splitext(os.path.basename(csv_file))[0] + ".csv"), encoding='utf-8')
# Apply concat_csv to all CSVs in cdir:
for f in excels:
concat_csv(os.path.join(cdir, f))
以下是Scott Boston推荐后使用的最终代码:
...
# Define concat_csv Function:
def concat_csv(csv_file):
df_csv = pd.read_csv(os.path.join(cdir, csv_file), header=None, index_col=None) # Load CSV into data frame
df = df_csv[df_csv.isin(["string 1", "string2"]).any(axis=1)] # Filter data frame by UGL data
df2 = df.dropna(axis=1, how="all") # Drop columns with all empty cells
try:
df_final = df2.set_index([0]) # Set index to column 1
except:
df_final = df2
# Send final dataframe to CSV in output directory:
df_final.to_csv(os.path.join(odir, os.path.splitext(os.path.basename(csv_file))[0] + ".csv"), encoding='utf-8')
# Apply concat_csv to all CSVs in cdir:
for f in excels:
concat_csv(os.path.join(cdir, f))
答案 0 :(得分:1)
IIUC:
您有一个包含N个系列的数据框,并且您想检查是否有任何系列中出现两个字符串,并构建一个只包含这些行的新数据框。
构建通用数据
df = pd.DataFrame({'A':np.random.choice(list('ABCDEFG'),size=26),'B':np.random.choice(list('FGHIJKLMN'),size=26)})
查找“G”或“F”出现在任何列中的所有记录
df_final = df[df.isin(['G','F']).any(axis=1)]
print(df_final)
输出:
A B
0 G I
2 G G
4 A G
7 F N
8 F M
10 C F
11 A G
14 F G
16 G H
18 F L
19 D G