技術標籤：python

pandas讀取檔案

starts=time.time()

for file in files[:1]:
    print(file)
    file_paths = os.path.join(root, file)
    print(file_paths)
    list2=[]

    # with open(file_paths,'r') as f:
    #    for row in f:
    #         list2 = [row.split(',')[line] for row in f]
    #         #list2.append(row.split()[0])
 

    #         app_data[file]=list2

    df_y = pd.read_csv(file_paths, engine='python')[data]

print(app_data)
end=time.time()

python open讀取檔案

line=int(data)
# x檔案有很多
for root, dirs, files in os.walk(train_file_x):
    # 用第一個檔案作為例子
    app_data = pd.DataFrame()

    starts=time.time()

    for file
 
 in files[:1]:
        print(file)
        file_paths = os.path.join(root, file)
        print(file_paths)
        list2=[]

        with open(file_paths,'r') as f:
           for row in f:
                list2 = [row.split(',')[line] for row in f]
                #list2.append(row.split()[0])
                app_data[
 
file]=list2

        #df_y = pd.read_csv(file_paths, engine='python')[data]

    print(app_data)

對比結果：
pandas和python：open 讀取同樣的檔案，花費的效率相差 5倍

所以，提高效率的辦法就是用 python：open讀取csv檔案，然後在轉成DataFrame，如果之後需要用DataFrame的話。