统计代写|风险建模代写Financial risk modeling代考|Valuing the economic benefit by simulations

如果你也在 怎样代写风险建模Financial risk modeling这个学科遇到相关的难题,请随时右上角联系我们的24/7代写客服。

风险建模是确定有多少风险存在于一个特定的企业、投资或一系列的现金流中。

statistics-lab™ 为您的留学生涯保驾护航 在代写风险建模Financial risk modeling方面已经树立了自己的口碑, 保证靠谱, 高质且原创的统计Statistics代写服务。我们的专家在代写风险建模Financial risk modeling代写方面经验极为丰富,各种代写风险建模Financial risk modeling相关的作业也就用不着说。

我们提供的风险建模Financial risk modeling及其相关学科的代写,服务范围广, 其中包括但不限于:

  • Statistical Inference 统计推断
  • Statistical Computing 统计计算
  • Advanced Probability Theory 高等楖率论
  • Advanced Mathematical Statistics 高等数理统计学
  • (Generalized) Linear Models 广义线性模型
  • Statistical Machine Learning 统计机器学习
  • Longitudinal Data Analysis 纵向数据分析
  • Foundations of Data Science 数据科学基础
Portfolio selection: a fuzzy-ANP approach | Financial Innovation | Full Text
统计代写|风险建模代写Financial risk modeling代考|Valuing the economic benefit by simulations

统计代写|风险建模代写Financial risk modeling代考|Valuing the economic benefit by simulations

In the following sections we show several numerical experiments to assess the gains offered by the Fourier estimator over other estimators in terms of in-sample and out-of-sample properties and from the

perspective of an asset-allocation decision problem. In Section 1.4.1 our attention is focused mainly on covariance estimation, since in this respect effects due to both nonsynchronicity and microstructure noise become effective. As for the finite sample variance analysis of the Fourier method, we refer the reader to Mancino and Sanfelici (2008a) for in-sample statistics and to Barucci et al. (2008) for the forecasting performance. Nevertheless, the results in Sections 1.4.2, 1.4.3 and 1.4.4 can be fully justified only by considering the properties of the different estimators for both the variance and the covariance measures. Following a large literature, we simulate discrete data from the continuous time bivariate Heston model
$$
\begin{aligned}
&d p^{1}(t)=\left(\mu_{1}-\sigma_{1}^{2}(t) / 2\right) d t+\sigma_{1}(t) d W_{1} \
&d p^{2}(t)=\left(\mu_{2}-\sigma_{2}^{2}(t) / 2\right) d t+\sigma_{2}(t) d W_{2} \
&d \sigma_{1}^{2}(t)=k_{1}\left(\alpha_{1}-\sigma_{1}^{2}(t)\right) d t+\gamma_{1} \sigma_{1}(t) d W_{3} \
&d \sigma_{2}^{2}(t)=k_{2}\left(\alpha_{2}-\sigma_{2}^{2}(t)\right) d t+\gamma_{2} \sigma_{2}(t) d W_{4}
\end{aligned}
$$
where $\operatorname{corr}\left(W_{1}, W_{2}\right)=0.35, \operatorname{corr}\left(W_{1}, W_{3}\right)=-0.5$ and $\operatorname{corr}\left(W_{2}, W_{4}\right)=$ $-0.55$. The other parameters of the model are as in Zhang et al. (2005): $\mu_{1}=0.05, \mu_{2}=0.055, k_{1}=5, k_{2}=5.5, \alpha_{1}=0.05, \alpha_{2}=0.045, \gamma_{1}=0.5$, $\gamma_{2}=0.5$. The volatility parameters satisfy Feller’s condition $2 k \alpha \geq \gamma^{2}$, which makes the zero boundary unattainable by the volatility process. Moreover, we assume that the additive logarithmic noises $\eta_{l}^{1}=\eta^{1}\left(t_{l}^{1}\right)$, $\eta_{l}^{2}=\eta^{2}\left(t_{l}^{2}\right)$ are i.i.d. Gaussian, contemporaneously correlated and independent from $p$. The correlation is set to $0.5$ and we assume $\omega_{i i}^{1 / 2}=$ $\left(E\left[\left(\eta^{i}\right)^{2}\right)^{1 / 2}=0,0.002,0.004\right.$, that is, we consider the case of no contamination and two different levels for the standard deviation of the noise. We also consider the case of dependent noise, assuming for simplicity $\eta_{l}^{i}=\alpha\left[p^{i}\left(t_{l}^{i}\right)-p^{i}\left(t_{l-1}^{i}\right)\right]+\bar{\eta}{l}^{i}$, for $i=1,2$ and $\bar{\eta}{l}^{i}$ i.i.d. Gaussian. We set $\alpha=0.1$. From the simulated data, integrated covariance estimates can be compared to the value of the true covariance quantities.

统计代写|风险建模代写Financial risk modeling代考|Covariance estimation and forecast

As a first application we perform an in-sample analysis in order to shed light on the properties of the different estimators in terms of different statistics of the covariance estimates, such as bias, MSE and others. More precisely, we consider the following relative error statistics:
$$
\mu=E\left[\frac{\hat{C}^{12}-\int_{0}^{2 \pi} \Sigma^{12}(t) d t}{\int_{0}^{2 \pi} \Sigma^{12}(t) d t}\right], \quad s t d=\left{\operatorname{Var}\left[\frac{\hat{C}^{12}-\int_{0}^{2 \pi} \Sigma^{12}(t) d t}{\int_{0}^{2 \pi} \Sigma^{12}(t) d t}\right]\right}^{1 / 2}
$$
which can be interpreted as relative bias and standard deviation of an estimator $\hat{C}^{12}$ for the covariance. The estimators have been optimized by choosing the cutting frequency $N$ of the Fourier expansion, the parameters $H$ and $S$ and the sampling interval for $R C^{o p t}$ on the basis of their MSE. The results are reported in Tables $1.1$ and $1.2$. Within each table, entries are the values of $\mu, s t d$, MSE and bias, using 750 Monte Carlo replications, which roughly correspond to three years. Rows correspond to the different estimators. The sampling interval for the realized covariance-type estimators is indicated as a superscript. The optimal sampling frequency for $R C$ opt is obtained by direct minimization of the true MSE of the covariance estimates and corresponds to $1 \mathrm{~min}$ in the absence of noise, to $1.33$ min when $\omega_{i i}^{1 / 2}=0.002$, to $1.67$ min when $\omega_{i i}^{1 / 2}=0.004$ and to $1.5 \mathrm{~min}$ when $\omega_{i i}^{1 / 2}=0.004$ and the noise is dependent on the price. The other optimal MSE-based parameter values are listed in the tables.

When we consider covariance estimates, the most important effect to deal with is the “Epps effect.” The presence of other microstructure effects represents a minor aspect in this respect. On the contrary, it may in some sense even compensate the effects due to nonsynchronicity, as we can see from the smaller MSE of the 1-min realized covariance estimator with respect to the 5 -min estimator in the cases with noise. We remark that the corresponding 1-min estimator for variances is more affected by the presence of noise, since it is not compensated for by nonsynchronicity. Moreover, in the absence of noise the Epps effect hampers consistency of the realized covariance estimates, yielding an optimal MSE-based frequency of $1 \mathrm{~min}$.

In fact, as with any estimator based on interpolated prices, the realized covariance-type estimators suffer from the Epps effect when trading is nonsynchronous. The lead-lag correction reduces such an effect, at least in terms of bias, to the disadvantage of a slightly larger MSE. Note that the lead-lag correction contrasts with the Epps effect, thus producing occasionally positive biases. In the absence of noise the best performance is achieved by the unbiased $\mathrm{AO}$ estimator and this justifies the optimal $S$ value for its subsampled version which is set to 1 , that is, no subsampling is needed. We remark that the optimal $H$ value for the kernel estimator $(K)$ is set to 4 , that is, the use of some weighted autocovariance is needed to contrast with the Epps effect, differently from the variance estimation, where the optimal MSE-based $H$ value is equal to 0 , which corresponds to the realized variance. On the other hand, the presence of noise strongly affects the AO estimator. This is due to the “Poisson trading scheme” with correlated noise. In fact, the $\mathrm{AO}$ remains unbiased under independent noise whenever the probability of trades occurring at the same time is zero, which is not the case for Poisson arrivals. In the same fashion, the Kernel estimator provides an acceptable estimate in the absence of noise but is rapidly swamped by the presence of noise. This is quite striking, because the corresponding variance estimator provides the best estimates at the highest frequencies in the presence of noise, as already discussed in Mancino and Sanfelici (2008a). Nevertheless, Barndorff-Nielsen et al.

统计代写|风险建模代写Financial risk modeling代考|Dynamic portfolio choice and economic gains

In this section, we consider the benefit of using the Fourier estimator with respect to others from the perspective of the asset-allocation problem of Section 1.3. Given any time series of daily variance/covariance estimates we split our samples of 750 days into two parts: the first one containing 30 percent of total estimates is used as a “burn-in” period, while the second one is saved for out-of-sample purposes. The out-of-sample forecast is based on univariate ARMA models, as in the previous section. More precisely, following Aït-Sahalia and Mancini (2008), the estimated series of 225 in-sample covariance matrices is used to fit univariate AR(1) models for each variance/covariance estimate separately.

The total number of out-of-sample forecasts $m$ for each series is equal to 525 . Each time a new forecast is performed, the corresponding actual variance/covariance measure is moved from the forecasting horizon to the first sample and the AR(1) parameters are reestimated in real time. Given sensible choices of $R f_{1} \mu_{p}$ and $\mu_{t}$, each one-day-ahead variance/covariance forecast leads to the determination of a daily

portfolio weight $\omega_{t}$. The time series of daily portfolio weights then leads to daily portfolio returns and utility estimation.

We implement the criterion in (1.14) by setting $R f$ equal to $0.03$ and considering three target $\mu_{p}$ values, namely $0.09,0.12$ and $0.15$. In order to concentrate on volatility timing and abstract from issues related to expected stock-return predictability, for all times $t$ we set the components of the vector $\mu_{t}=E_{t}\left[R_{t+1}\right]$ equal to the sample means of the returns on the risky assets over the forecasting horizon. For all times $t$, the conditional covariance matrix is computed as an out-of-sample forecast based on the different variance/covariance estimates.

We interpret the difference $U^{\hat{C}}-U^{\text {Fourier }}$ between the average utility computed on the basis of the Fourier estimator and that based on alternative estimators $\hat{C}$, as the fee that the investor would be willing to pay to switch from covariance forecasts based on estimator $\hat{C}$ to covariance forecasts based on the Fourier estimator. Tables 1.7, $1.8$ and $1.9$ contain the results for three levels of risk aversion and three target expected returns in the different noise scenarios considered in our analysis. We remark that, in general, the optimal sampling frequencies for the realized variances and covariances are different within each scenario, due to the different effects of microstructure noise and nonsynchronicity on the volatility measures. Therefore, in the asset-allocation application we chose to use a unique sampling frequency for realized variances and covariances, given by the maximum among the optimal sampling intervals corresponding to variances and covariances.

Multi-Period Portfolio Optimization with Investor Views under Regime  Switching
统计代写|风险建模代写Financial risk modeling代考|Valuing the economic benefit by simulations

风险建模代写

统计代写|风险建模代写Financial risk modeling代考|Valuing the economic benefit by simulations

在接下来的部分中,我们展示了几个数值实验来评估傅立叶估计器在样本内和样本外属性方面相对于其他估计器的增益,以及从

资产配置决策问题的视角。在 1.4.1 节中,我们的注意力主要集中在协方差估计上,因为在这方面,由于非同步性和微观结构噪声造成的影响变得有效。至于傅立叶方法的有限样本方差分析,我们请读者参考 Mancino 和 Sanfelici (2008a) 的样本内统计数据以及 Barucci 等人的文章。(2008)的预测性能。然而,第 1.4.2、1.4.3 和 1.4.4 节中的结果只有通过考虑方差和协方差测量的不同估计量的性质才能得到充分证明。根据大量文献,我们模拟了来自连续时间双变量 Heston 模型的离散数据
dp1(吨)=(μ1−σ12(吨)/2)d吨+σ1(吨)d在1 dp2(吨)=(μ2−σ22(吨)/2)d吨+σ2(吨)d在2 dσ12(吨)=ķ1(一种1−σ12(吨))d吨+C1σ1(吨)d在3 dσ22(吨)=ķ2(一种2−σ22(吨))d吨+C2σ2(吨)d在4
在哪里更正⁡(在1,在2)=0.35,更正⁡(在1,在3)=−0.5和更正⁡(在2,在4)= −0.55. 该模型的其他参数与 Zhang 等人的相同。(2005):μ1=0.05,μ2=0.055,ķ1=5,ķ2=5.5,一种1=0.05,一种2=0.045,C1=0.5, C2=0.5. 波动率参数满足 Feller 条件2ķ一种≥C2,这使得波动过程无法达到零边界。此外,我们假设加性对数噪声这l1=这1(吨l1),这l2=这2(吨l2)是独立高斯分布的,同时相关且独立于p. 相关性设置为0.5我们假设ω一世一世1/2= (和[(这一世)2)1/2=0,0.002,0.004,即我们考虑无污染的情况,噪声的标准差有两个不同的水平。我们还考虑了相关噪声的情况,为简单起见假设这l一世=一种[p一世(吨l一世)−p一世(吨l−1一世)]+这¯l一世, 为了一世=1,2和这¯l一世iid 高斯。我们设置一种=0.1. 从模拟数据中,可以将综合协方差估计值与真实协方差量的值进行比较。

统计代写|风险建模代写Financial risk modeling代考|Covariance estimation and forecast

作为第一个应用程序,我们执行样本内分析,以便根据协方差估计的不同统计数据(例如偏差、MSE 等)阐明不同估计量的属性。更准确地说,我们考虑以下相对误差统计:
\mu=E\left[\frac{\hat{C}^{12}-\int_{0}^{2 \pi} \Sigma^{12}(t) d t}{\int_{0}^{ 2 \pi} \Sigma^{12}(t) d t}\right], \quad s t d=\left{\operatorname{Var}\left[\frac{\hat{C}^{12}-\int_{ 0}^{2 \pi} \Sigma^{12}(t) d t}{\int_{0}^{2 \pi} \Sigma^{12}(t) d t}\right]\right}^{ 1 / 2}\mu=E\left[\frac{\hat{C}^{12}-\int_{0}^{2 \pi} \Sigma^{12}(t) d t}{\int_{0}^{ 2 \pi} \Sigma^{12}(t) d t}\right], \quad s t d=\left{\operatorname{Var}\left[\frac{\hat{C}^{12}-\int_{ 0}^{2 \pi} \Sigma^{12}(t) d t}{\int_{0}^{2 \pi} \Sigma^{12}(t) d t}\right]\right}^{ 1 / 2}
这可以解释为估计量的相对偏差和标准偏差C^12为协方差。通过选择切割频率对估计器进行了优化ñ傅里叶展开的参数H和小号和采样间隔RC这p吨基于他们的MSE。结果报告在表中1.1和1.2. 在每个表中,条目是μ,s吨d、MSE 和偏差,使用 750 次蒙特卡洛复制,大致相当于三年。行对应于不同的估计量。已实现协方差类型估计器的采样间隔用上标表示。最佳采样频率RCopt 是通过直接最小化协方差估计的真实 MSE 获得的,对应于1 米一世n在没有噪音的情况下,1.33最小时间ω一世一世1/2=0.002, 到1.67最小时间ω一世一世1/2=0.004并1.5 米一世n什么时候ω一世一世1/2=0.004噪音取决于价格。表中列出了其他基于 MSE 的最佳参数值。

当我们考虑协方差估计时,要处理的最重要的影响是“埃普斯效应”。其他微观结构效应的存在代表了这方面的一个次要方面。相反,它在某种意义上甚至可以补偿由于非同步性造成的影响,正如我们在有噪声的情况下从 1 分钟实现协方差估计器相对于 5 分钟估计器的较小 MSE 中可以看出的那样。我们注意到相应的 1 分钟方差估计量更受噪声存在的影响,因为它没有被非同步性补偿。此外,在没有噪声的情况下,Epps 效应会阻碍实现的协方差估计的一致性,从而产生基于 MSE 的最佳频率1 米一世n.

事实上,与任何基于插值价格的估计器一样,当交易不同步时,已实现的协方差型估计器会受到 Epps 效应的影响。至少在偏差方面,超前滞后校正减少了这种影响,从而导致 MSE 稍大的缺点。请注意,领先滞后校正与 Epps 效应形成对比,因此偶尔会产生正偏差。在没有噪声的情况下,最佳性能是由无偏的一种这估计器,这证明了最优小号其子采样版本的值设置为 1 ,即不需要子采样。我们注意到最优H内核估计器的值(ķ)设置为 4 ,即需要使用一些加权自协方差来与 Epps 效应进行对比,这与方差估计不同,其中基于 MSE 的最优Hvalue 等于 0 ,对应于已实现的方差。另一方面,噪声的存在强烈影响 AO 估计器。这是由于具有相关噪声的“泊松交易方案”。事实上,一种这当交易同时发生的概率为零时,在独立噪声下保持无偏,泊松到达的情况并非如此。以同样的方式,内核估计器在没有噪声的情况下提供可接受的估计,但很快就会被噪声的存在淹没。这是相当惊人的,因为相应的方差估计器在存在噪声的情况下提供了最高频率的最佳估计,正如 Mancino 和 Sanfelici (2008a) 中已经讨论的那样。尽管如此,Barndorff-Nielsen 等人。

统计代写|风险建模代写Financial risk modeling代考|Dynamic portfolio choice and economic gains

在本节中,我们从 1.3 节的资产配置问题的角度考虑使用傅立叶估计量相对于其他估计量的好处。给定每日方差/协方差估计的任何时间序列,我们将 750 天的样本分成两部分:第一个包含 30% 的总估计值用作“老化”期,而第二个保留用于输出样本目的。如上一节所述,样本外预测基于单变量 ARMA 模型。更准确地说,根据 Aït-Sahalia 和 Mancini (2008),估计的 225 个样本内协方差矩阵系列用于分别拟合每个方差/协方差估计的单变量 AR(1) 模型。

样本外预测总数米每个系列等于 525 。每次执行新的预测时,相应的实际方差/协方差度量从预测范围移动到第一个样本,并且实时重新估计 AR(1) 参数。鉴于明智的选择RF1μp和μ吨,每个前一天的方差/协方差预测导致确定每日

投资组合权重ω吨. 然后,每日投资组合权重的时间序列导致每日投资组合回报和效用估计。

我们通过设置来实现(1.14)中的标准RF等于0.03并考虑三个目标μp值,即0.09,0.12和0.15. 为了专注于波动时间并从与预期股票收益可预测性相关的问题中抽象出来,始终吨我们设置向量的分量μ吨=和吨[R吨+1]等于预测期内风险资产收益的样本均值。一直以来吨,条件协方差矩阵计算为基于不同方差/协方差估计的样本外预测。

我们解释差异在C^−在傅立叶 基于傅立叶估计器计算的平均效用与基于替代估计器计算的平均效用之间的关系C^,作为投资者愿意支付的费用,以从基于估计量的协方差预测转换C^基于傅里叶估计的协方差预测。表 1.7,1.8和1.9包含我们分析中考虑的不同噪声情景中三个风险厌恶程度和三个目标预期回报的结果。我们注意到,一般来说,由于微观结构噪声和非同步性对波动率测量的影响不同,实现方差和协方差的最佳采样频率在每种情况下都是不同的。因此,在资产配置应用程序中,我们选择对已实现的方差和协方差使用唯一的采样频率,由对应于方差和协方差的最佳采样间隔中的最大值给出。

统计代写|风险建模代写Financial risk modeling代考 请认准statistics-lab™

统计代写请认准statistics-lab™. statistics-lab™为您的留学生涯保驾护航。

金融工程代写

金融工程是使用数学技术来解决金融问题。金融工程使用计算机科学、统计学、经济学和应用数学领域的工具和知识来解决当前的金融问题,以及设计新的和创新的金融产品。

非参数统计代写

非参数统计指的是一种统计方法,其中不假设数据来自于由少数参数决定的规定模型;这种模型的例子包括正态分布模型和线性回归模型。

广义线性模型代考

广义线性模型(GLM)归属统计学领域,是一种应用灵活的线性回归模型。该模型允许因变量的偏差分布有除了正态分布之外的其它分布。

术语 广义线性模型(GLM)通常是指给定连续和/或分类预测因素的连续响应变量的常规线性回归模型。它包括多元线性回归,以及方差分析和方差分析(仅含固定效应)。

有限元方法代写

有限元方法(FEM)是一种流行的方法,用于数值解决工程和数学建模中出现的微分方程。典型的问题领域包括结构分析、传热、流体流动、质量运输和电磁势等传统领域。

有限元是一种通用的数值方法,用于解决两个或三个空间变量的偏微分方程(即一些边界值问题)。为了解决一个问题,有限元将一个大系统细分为更小、更简单的部分,称为有限元。这是通过在空间维度上的特定空间离散化来实现的,它是通过构建对象的网格来实现的:用于求解的数值域,它有有限数量的点。边界值问题的有限元方法表述最终导致一个代数方程组。该方法在域上对未知函数进行逼近。[1] 然后将模拟这些有限元的简单方程组合成一个更大的方程系统,以模拟整个问题。然后,有限元通过变化微积分使相关的误差函数最小化来逼近一个解决方案。

tatistics-lab作为专业的留学生服务机构,多年来已为美国、英国、加拿大、澳洲等留学热门地的学生提供专业的学术服务,包括但不限于Essay代写,Assignment代写,Dissertation代写,Report代写,小组作业代写,Proposal代写,Paper代写,Presentation代写,计算机作业代写,论文修改和润色,网课代做,exam代考等等。写作范围涵盖高中,本科,研究生等海外留学全阶段,辐射金融,经济学,会计学,审计学,管理学等全球99%专业科目。写作团队既有专业英语母语作者,也有海外名校硕博留学生,每位写作老师都拥有过硬的语言能力,专业的学科背景和学术写作经验。我们承诺100%原创,100%专业,100%准时,100%满意。

随机分析代写


随机微积分是数学的一个分支,对随机过程进行操作。它允许为随机过程的积分定义一个关于随机过程的一致的积分理论。这个领域是由日本数学家伊藤清在第二次世界大战期间创建并开始的。

时间序列分析代写

随机过程,是依赖于参数的一组随机变量的全体,参数通常是时间。 随机变量是随机现象的数量表现,其时间序列是一组按照时间发生先后顺序进行排列的数据点序列。通常一组时间序列的时间间隔为一恒定值(如1秒,5分钟,12小时,7天,1年),因此时间序列可以作为离散时间数据进行分析处理。研究时间序列数据的意义在于现实中,往往需要研究某个事物其随时间发展变化的规律。这就需要通过研究该事物过去发展的历史记录,以得到其自身发展的规律。

回归分析代写

多元回归分析渐进(Multiple Regression Analysis Asymptotics)属于计量经济学领域,主要是一种数学上的统计分析方法,可以分析复杂情况下各影响因素的数学关系,在自然科学、社会和经济学等多个领域内应用广泛。

MATLAB代写

MATLAB 是一种用于技术计算的高性能语言。它将计算、可视化和编程集成在一个易于使用的环境中,其中问题和解决方案以熟悉的数学符号表示。典型用途包括:数学和计算算法开发建模、仿真和原型制作数据分析、探索和可视化科学和工程图形应用程序开发,包括图形用户界面构建MATLAB 是一个交互式系统,其基本数据元素是一个不需要维度的数组。这使您可以解决许多技术计算问题,尤其是那些具有矩阵和向量公式的问题,而只需用 C 或 Fortran 等标量非交互式语言编写程序所需的时间的一小部分。MATLAB 名称代表矩阵实验室。MATLAB 最初的编写目的是提供对由 LINPACK 和 EISPACK 项目开发的矩阵软件的轻松访问,这两个项目共同代表了矩阵计算软件的最新技术。MATLAB 经过多年的发展,得到了许多用户的投入。在大学环境中,它是数学、工程和科学入门和高级课程的标准教学工具。在工业领域,MATLAB 是高效研究、开发和分析的首选工具。MATLAB 具有一系列称为工具箱的特定于应用程序的解决方案。对于大多数 MATLAB 用户来说非常重要,工具箱允许您学习应用专业技术。工具箱是 MATLAB 函数(M 文件)的综合集合,可扩展 MATLAB 环境以解决特定类别的问题。可用工具箱的领域包括信号处理、控制系统、神经网络、模糊逻辑、小波、仿真等。

R语言代写问卷设计与分析代写
PYTHON代写回归分析与线性模型代写
MATLAB代写方差分析与试验设计代写
STATA代写机器学习/统计学习代写
SPSS代写计量经济学代写
EVIEWS代写时间序列分析代写
EXCEL代写深度学习代写
SQL代写各种数据建模与可视化代写

发表回复

您的电子邮箱地址不会被公开。 必填项已用 * 标注