<span>#</span><span> -*- coding: utf-8 -*-</span>
<span>"""</span><span>
Spyder Editor

This is a temporary script file.
</span><span>"""</span>

<span><a href="https://www.gaodaima.com/tag/import" title="查看更多关于import的文章" target="_blank">import</a></span><span> pandas as pd
</span><span>import</span><span> pandas_datareader
</span><span>import</span><span> datetime
</span><span>import</span><span> matplotlib.pylab as plt
</span><span>from</span> matplotlib.pylab <span>import</span><span> style
</span><span>from</span> statsmodels.tsa.arima_model <span>import</span><span> ARIMA
</span><span>from</span> statsmodels.graphics.tsaplots <span>import</span><span> plot_acf, plot_pacf

style.use(</span><span>"</span><span>ggplot</span><span>"</span>)     <span>#</span><span> 设置图片显示的主题样式</span>

<span>#</span><span> 解决matplotlib显示中文问题</span>
plt.rcParams[<span>"</span><span>font.sans-serif</span><span>"</span>] = [<span>"</span><span>SimHei</span><span>"</span>]  <span>#</span><span> 指定默认字体</span>
plt.rcParams[<span>"</span><span>axes.unicode_minus</span><span>"</span>] = False  <span>#</span><span> 解决保存图像是负号"-"显示为方块的问题</span>


<span>def</span><span> run_main():
    </span><span>"""</span><span>
        主函数
    </span><span>"""</span>
    <span>#</span><span> 1. 准备数据</span>
    <span>#</span><span> 指定股票分析开始日期</span>
    start_date = datetime.datetime(2009, 1, 1<span>)
    </span><span>#</span><span> 指定股票分析截止日期</span>
    end_date = datetime.datetime(2019, 4, 1<span>)
    </span><span>#</span><span> 股票代码</span>
    stock_code = <span>"</span><span>600519.SS</span><span>"</span>    <span>#</span><span> 沪市贵州茅台</span>
<span>
    stock_df </span>=<span> pandas_datareader.data.DataReader(
                        stock_code, </span><span>"</span><span>yahoo</span><span>"</span><span>, start_date, end_date
                )
    </span><span>#</span><span> 预览数据</span>
    <span>print</span><span>(stock_df.head())

    </span><span>#</span><span> 2. 可视化数据</span>
    plt.plot(stock_df[<span>"</span><span>Close</span><span>"</span><span>])
    plt.title(</span><span>"</span><span>股票每日收盘价</span><span>"</span><span>)
    plt.show()

    </span><span>#</span><span> 按周重采样</span>
    stock_s = stock_df[<span>"</span><span>Close</span><span>"</span>].resample(<span>"</span><span>W-MON</span><span>"</span><span>).mean()
    stock_train </span>= stock_s[<span>"</span><span>2014</span><span>"</span>:<span>"</span><span>2018</span><span>"</span><span>]
    plt.plot(stock_train)
    plt.title(</span><span>"</span><span>股票周收盘价均值</span><span>"</span><span>)
    plt.show()

    </span><span>#</span><span> 分析 ACF</span>
    acf = plot_acf(stock_train, lags=20<span>)
    plt.title(</span><span>"</span><span>股票指数的 ACF</span><span>"</span><span>)
    acf.show()

    </span><span>#</span><span> 分析 PACF</span>
    pacf = plot_pacf(stock_train, lags=20<span>)
    plt.title(</span><span>"</span><span>股票指数的 PACF</span><span>"</span><span>)
    pacf.show()

    </span><span>#</span><span> 3. 处理数据，平稳化数据</span>
    <span>#</span><span> 这里只是简单第做了一节差分，还有其他平稳化时间序列的方法</span>
    stock_diff =<span> stock_train.diff()
    diff </span>=<span> stock_diff.dropna()
    </span><span>print</span><span>(diff.head())
    </span><span>print</span><span>(diff.dtypes)

    plt.figure()
    plt.plot(diff)
    plt.title(</span><span>"</span><span>一阶差分</span><span>"</span><span>)
    plt.show()

    acf_diff </span>= plot_acf(diff, lags=20<span>)
    plt.title(</span><span>"</span><span>一阶差分的 ACF</span><span>"</span><span>)
    acf_diff.show()

    pacf_diff </span>= plot_pacf(diff, lags=20<span>)
    plt.title(</span><span>"</span><span>一阶差分的 PACF</span><span>"</span><span>)
    pacf_diff.show()

    </span><span>#</span><span> 4. 根据ACF和PACF定阶并建立模型</span>
    model = ARIMA(stock_train, order=(1, 1, 1), freq=<span>"</span><span>W-MON</span><span>"</span><span>)
    </span><span>#</span><span> 拟合模型</span>
    arima_result =<span> model.fit()
    </span><span>print</span><span>(arima_result.summary())

    </span><span>#</span><span> 5. 预测</span>
<span>
    pred_vals </span>= arima_result.predict(start=str(<span>"</span><span>2019-01</span><span>"</span>),end=str(<span>"</span><span>2019-03</span><span>"</span><span>),
                                     dynamic</span>=False, typ=<span>"</span><span>levels</span><span>"</span><span>)
    </span><span>print</span><span>(pred_vals)

    </span><span>#</span><span> 6. 可视化预测结果</span>
    stock_forcast = pd.concat([stock_s, pred_vals], axis=1, keys=[<span>"</span><span>original</span><span>"</span>, <span>"</span><span>predicted</span><span>"</span><span>])

    plt.figure()
    plt.plot(stock_forcast)
    plt.title(</span><span>"</span><span>真实值vs预测值</span><span>"</span><span>)
    plt.savefig(</span><span>"</span><span>./stock_pred.png</span><span>"</span>, format=<span>"</span><span>png</span><span>"</span><span>)
    plt.show()


</span><span>if</span> <span>__name__</span> == <span>"</span><span>__main__</span><span>"</span><span>:
    run_main()</span>

www#gaodaima.com来源gaodai$ma#com搞$$代**码)网搞代码