例如，我有以下格式的數(shù)據(jù)文本：HEADER NODE DATA, AIR -10000, 15., -1.0HEADER CONDUCTOR DATA, AIR 1, AIR.10000, S25D.1, 56.84441 $ AIR CONV 2, AIR.10000, S25D.2, 56.45712 $ AIR CONV 3, AIR.10000, S25D.3, 53.35623 $ AIR CONV 4, AIR.10000, S25D.4, 45.09633 $ AIR CONV 5, AIR.10000, S25D.9, 77.00067 $ AIR CONV 6, AIR.10000, S25D.10, 80.35013 $ AIR CONV 7, AIR.10000, S25D.11, 50.4933 $ AIR CONV 8, AIR.10000, S25D.12, 91.61026 $ AIR CONV 9, AIR.10000, S25D.13, 53.75025 $ AIR CONV 10, AIR.10000, S25D.14, 75.68577 $ AIR CONV 11, AIR.10000, S25D.15, 110.0111 $ AIR CONV 12, AIR.10000, S25D.16, 114.7913 $ AIR CONV 13, AIR.10000, S25D.17, 81.12207 $ AIR CONV 14, AIR.10000, S25D.18, 72.80061 $ AIR CONV 15, AIR.10000, S25D.19, 72.21327 $ AIR CONV 16, AIR.10000, S25D.20, 90.99183 $ AIR CONV 17, AIR.10000, S25D.21, 66.35648 $ AIR CONV 18, AIR.10000, S25D.22, 76.9787 $ AIR CONV 19, AIR.10000, S25D.23, 52.46601 $ AIR CONV 20, AIR.10000, S25D.24, 68.30105 $ AIR CONV 21, AIR.10000, S25D.25, 114.0903 $ AIR CONV 22, AIR.10000, S25D.26, 70.51425 $ AIR CONV 23, AIR.10000, S25D.27, 36.39104 $ AIR CONV我想將其讀入熊貓數(shù)據(jù)框以進(jìn)行進(jìn)一步分析。如您所見，每個(gè)子集的標(biāo)題和數(shù)據(jù)格式都不同。我想使用 re 模塊和 pandas 將數(shù)據(jù)讀入框架，但不知道如何以最有效的方式進(jìn)行。此外，您可以看到 AIR 字符串對兩個(gè)標(biāo)頭都是通用的。在完整文件中會有更多這樣的字符串，所以我想為每個(gè)字符串創(chuàng)建一個(gè)類，其屬性（節(jié)點(diǎn)，導(dǎo)體）將是一個(gè)帶有相應(yīng)數(shù)據(jù)的 pd 數(shù)據(jù)框。
1 回答
12345678_0001
TA貢獻(xiàn)1802條經(jīng)驗(yàn) 獲得超5個(gè)贊
您可以嘗試以下方法：
使用逐行讀取文件
f.readlines()對于每一行，檢查該行是否包含
HEADER.如果是這種情況，這意味著這是一個(gè)新的子 df。將其添加到輸出
else: 將當(dāng)前行追加到當(dāng)前子數(shù)據(jù)
將文本格式的 df 列表轉(zhuǎn)換
sub為 pandas 數(shù)據(jù)框。這個(gè)討論解釋了如何。筆記：將任何“
,”與“,*”匹配使用匹配任何所有2 個(gè)或更多空間
"\s{2,}使用的正則表達(dá)式是
,*\s{2,}指：該參數(shù)
header=None是必需的，第一行將用于添加列名稱。
完整代碼
# import modules
import pandas as pd
from io import StringIO
out = []
with open("data.txt", "r") as f:
sub = ""
for line in f.readlines():
if "HEADER" in line:
# Skip first insertion
if sub != "":
out.append(sub)
sub = ""
else: sub += line
# Append last sub (outside for loop)
out.append(sub)
# "out" is a list of pandas dataframes
out = [pd.read_csv(StringIO(o), sep=r",*\s{2,}", header=None, engine="python") for o in out]
print(out[0])
# 0 1 2
# 0 -10000 15.0 -1.0
print(out[1])
# 0 1 2 3 4
# 0 1 AIR.10000 S25D.1 56.84441 $ AIR CONV
# 1 2 AIR.10000 S25D.2 56.45712 $ AIR CONV
# 2 3 AIR.10000 S25D.3 53.35623 $ AIR CONV
# 3 4 AIR.10000 S25D.4 45.09633 $ AIR CONV
# 4 5 AIR.10000 S25D.9 77.00067 $ AIR CONV
# .. .. ... ... ... ...
# 89 90 AIR.10000 S25D.122 133.79200 $ AIR CONV
# 90 91 AIR.10000 S25D.123 194.22270 $ AIR CONV
# 91 92 AIR.10000 S25D.124 116.83720 $ AIR CONV
# 92 93 AIR.10000 S25D.125 125.51610 $ AIR CONV
# 93 94 AIR.10000 S25D.126 54.22872 $ AIR CONV
# [94 rows x 5 columns]
print(out[1].columns)
# Int64Index([0, 1, 2, 3, 4], dtype='int64')
添加回答
舉報(bào)
0/150
提交
取消