多列的测试训练数据框

Question

我有一个 csv 文件

Date,Open,High,Low,Close,Adj Close,Volume,Cash EPS,Book Value,Div/share,Net profit/share,NPM,ROE,ROCE,ROA,DEBT/EQ,ATR,CR
2004-04-26,82.924217,82.924217,82.924217,82.924217,60.026066,0,221.24,488.21,129.5,186.6,26.11,38.22,38.22,24.2,0,92.67,1.65
2004-04-27,82.778122,82.778122,79.765625,80.24453,58.086323,28616000,221.24,488.21,129.5,186.6,26.11,38.22,38.22,24.2,0,92.67,1.65

只给出 2 行以便于计算。我创建了一个数据框

import pandas as pd
import matplotlib.pyplot as plt
import numpy as np
from sklearn.preprocessing import MinMaxScaler

dataframe1 = pd.read_csv('test.csv')
df = dataframe1.dropna()
scaler=MinMaxScaler(feature_range=(0,1))
df1=scaler.fit_transform(np.array(df1).reshape(-1,1))
min_max_scaler = MinMaxScaler()
df[["Open", "High", "Low", "Close", "Adj Close", "Volume", "Book Value", "Div/share", "Net profit/share", "NPM", "ROE", "ROCE", "ROA", "DEBT/EQ", "ATR", "CR"]] = min_max_scaler.fit_transform(df[["Open", "High", "Low", "Close", "Adj Close", "Volume", "Book Value", "Div/share", "Net profit/share", "NPM", "ROE", "ROCE", "ROA", "DEBT/EQ", "ATR", "CR"]])

要训练数据集，我需要日期和预测，即关闭列。但是，关闭列值取决于多个列（即此 csv 中存在的所有列）

如何基于所有其他列训练日期和关闭列的数据，以便可以预测未来的关闭？

Answer 1

如果我理解这个问题，那么您正在寻找多变量时间序列模型。换句话说，它需要为每个时间步长输入多个变量才能做出前瞻性预测。下面是一些示例 link：

https://www.relataly.com/stock-market-prediction-with-multivariate-time-series-in-python/1815/

此外，我建议查看 Kaggle 股票市场预测竞赛，这里有数百个关于人们如何解决这个问题的例子。

https://www.kaggle.com/c/two-sigma-financial-news

多列的测试训练数据框

Test Train Dataframe for multiple columns

python

numpy

pandas

scikit-learn

data-science