Python基于决策树算法预测黄金价格走势

!pip install pyecharts -i https://pypi.mirrors.ustc.edu.cn/simple

import pandas as pd
from pyecharts.charts import *
from pyecharts.charts import Bar
from pyecharts.charts import Map
from pyecharts.charts import Pie
from pyecharts.charts import Line
from pyecharts.charts import Timeline
from pyecharts import options as opts
import warnings
warnings.filterwarnings('ignore')

#读取数据
df = pd.read_csv('/home/mw/input/data2152/gold_data.csv')

df.head(10)

df.tail(10)

df.info()

<class 'pandas.core.frame.DataFrame'>
RangeIndex: 2547 entries, 0 to 2546
Data columns (total 6 columns):
 #   Column      Non-Null Count  Dtype  
---  ------      --------------  -----  
 0   Date        2547 non-null   object 
 1   Close/Last  2547 non-null   float64
 2   Volume      2508 non-null   float64
 3   Open        2547 non-null   float64
 4   High        2547 non-null   float64
 5   Low         2547 non-null   float64
dtypes: float64(5), object(1)
memory usage: 119.5+ KB

#更改时间日期格式
df['Date'] = pd.to_datetime(df['Date'], format="%m/%d/%Y")
df.head()

#返回每一列缺失值统计

print(df.isnull().sum())
df['Volume'].fillna(method='bfill', inplace=True) # 使用前一个有效值填补
print(df.isnull().sum())

Date           0
Close/Last     0
Volume        39
Open           0
High           0
Low            0
dtype: int64
Date          0
Close/Last    0
Volume        0
Open          0
High          0
Low           0
dtype: int64

df.duplicated().sum()

0

#k线图
# 提取需要用于绘图的数据
df['Date'] = pd.to_datetime(df['Date'])
date = df['Date']
open_price = df['Open']
close_price = df['Close/Last']
high_price = df['High']
low_price = df['Low']

# 设置日期范围
start_date = '2012-10-28'
end_date = '2022-10-28'

df_filtered = df[(df['Date'] >= start_date) & (df['Date'] <= end_date)]
df_filtered = df_filtered.sort_values('Date', ascending=True)  # 按日期升序排列
kline = (
    Kline()
    .add_xaxis(df_filtered['Date'].tolist())
    .add_yaxis('K线图', [[o, c, l, h] for o, c, l, h in zip(df_filtered['Open'], df_filtered['Close/Last'], df_filtered['Low'], df_filtered['High'])])
    .set_global_opts(
        title_opts=opts.TitleOpts(title="黄金价格k线图"),
        xaxis_opts=opts.AxisOpts(type_="category"),
        yaxis_opts=opts.AxisOpts(name="价格"),
    )
)

kline.render_notebook()

# 创建Bar对象
bar = (
    Bar()
    .add_xaxis(df_filtered['Date'].tolist())
    .add_yaxis("成交量", df_filtered['Volume'].tolist())
    .set_global_opts(
        xaxis_opts=opts.AxisOpts(
            type_="category",
            axislabel_opts=opts.LabelOpts(rotate=-45),
            is_scale=True,  # 开启拖拽缩放
            boundary_gap=True,  # 设置坐标轴两端留白
        ),
        yaxis_opts=opts.AxisOpts(name="成交量"),
        title_opts=opts.TitleOpts(title="黄金价格成交量柱状图"),
        toolbox_opts=opts.ToolboxOpts(is_show=True),  # 显示工具栏
        datazoom_opts=[opts.DataZoomOpts(), opts.DataZoomOpts(type_="inside")],  # 添加数据缩放组件
    )
)
bar.render_notebook()

# 计算移动平均线
ma5 = df_filtered['Close/Last'].rolling(window=5).mean()
ma20 = df_filtered['Close/Last'].rolling(window=20).mean()

# 将移动平均线数据添加到DataFrame
df_filtered['MA5'] = ma5
df_filtered['MA20'] = ma20
# 创建Line对象
line = (
    Line()
    .add_xaxis(df_filtered['Date'].tolist())
    .add_yaxis("收盘价", df_filtered['Close/Last'].tolist())
    .add_yaxis("5日移动平均线", df_filtered['MA5'].tolist())
    .add_yaxis("20日移动平均线", df_filtered['MA20'].tolist())
    .set_global_opts(
        xaxis_opts=opts.AxisOpts(
            type_="category",
            axislabel_opts=opts.LabelOpts(rotate=-45),
            is_scale=True,  # 开启拖拽缩放
            boundary_gap=True,  # 设置坐标轴两端留白
        ),
        yaxis_opts=opts.AxisOpts(name="价格"),
        title_opts=opts.TitleOpts(title="黄金价格移动平均线图"),
        toolbox_opts=opts.ToolboxOpts(is_show=True),  # 显示工具栏
        datazoom_opts=[opts.DataZoomOpts(), opts.DataZoomOpts(type_="inside")],  # 添加数据缩放组件
    )
)
line.render_notebook()

# 筛选出2012-2018年的数据作为训练集
start_date = '2012-10-28'
end_date = '2019-12-31'
train_df = df_filtered[(df_filtered['Date'] >= start_date) & (df_filtered['Date'] <= end_date)]

# 剩下的数据作为测试集
test_df = df_filtered[df_filtered['Date'] > end_date]
train_df.shape,test_df.shape

((1834, 8), (713, 8))

df_filtered

from pycaret.regression import *
#创建回归模型
reg1 = setup(data = train_df,
            session_id=23,
            target = 'Close/Last',                      #指定目标列
            categorical_features=["MA5","MA20"],          #显式指定类别量的行
            feature_selection=True,feature_selection_threshold=0.8,    #去除次要特征
            normalize = True
            )

Setup Succesfully Completed.

#预选模型分别为决策树和三个梯度提升决策树模型
compare_models(sort='RMSE',whitelist=['dt','xgboost','lightgbm','catboost'])

LGBMRegressor(boosting_type='gbdt', class_weight=None, colsample_bytree=1.0,
              importance_type='split', learning_rate=0.1, max_depth=-1,
              min_child_samples=20, min_child_weight=0.001, min_split_gain=0.0,
              n_estimators=100, n_jobs=-1, num_leaves=31, objective=None,
              random_state=23, reg_alpha=0.0, reg_lambda=0.0, silent=True,
              subsample=1.0, subsample_for_bin=200000, subsample_freq=0)

#dt评分是最高，因此选择创建一个dt模型
LGBM = create_model("lightgbm")

# 通过tune_model处理模型对象，对其超参数进行调整
tuned_LGBM = tune_model(LGBM)

# 在之前划分的数据集上进行预测

predictions = predict_model(tuned_LGBM, data = test_df)predictions

# 提取"Date"、"Close/Last"和"Label"列的数据
dates = predictions['Date'].tolist()
close_last = predictions['Close/Last'].tolist()
label = predictions['Label'].tolist()

# 创建折线图对象
line_chart = Line()

# 添加"Close/Last"数据线
line_chart.add_xaxis(dates)
line_chart.add_yaxis("Close/Last", close_last)

# 添加"Label"数据线
line_chart.add_yaxis("Label", label)

# 设置全局配置
line_chart.set_global_opts(
    title_opts=opts.TitleOpts(title="Close/Last and Label Line Chart"),
    xaxis_opts=opts.AxisOpts(axislabel_opts=opts.LabelOpts(rotate=-45)),
    yaxis_opts=opts.AxisOpts(name='Value'),
)

# 渲染图表并保存为html文件
line_chart.render_notebook()

import numpy as np
label = np.array(predictions["Label"])
target = np.array(test_df["Close/Last"])

error = label - target
# 计算均方根误差RMSE
RMSE = np.nanmean((error**2))**0.5
# 计算平均绝对误差MAE 
MAE = np.nanmean(abs(error))
print('RMSE:',RMSE)
print('MAE:',MAE)

RMSE: 122.60895426490625
MAE: 95.08966605890605