分数阶微分_分数微分特征

分数阶微分

📈Python金融系列 (📈Python for finance series)

Warning: There is no magical formula or Holy Grail here, though a new world might open the door for you.

警告 ： 这里没有神奇的配方或圣杯，尽管新世界可能为您打开了大门。

📈Python金融系列 (📈Python for finance series)

1. Identifying Outliers

   识别异常值

2. Identifying Outliers — Part Two

   识别异常值-第二部分

3. Identifying Outliers — Part Three

   识别异常值-第三部分

4. Stylized Facts

   程式化的事实

5. Feature Engineering & Feature Selection

   特征工程与特征选择

6. Data Transformation

   数据转换

7. Fractionally Differentiated Features

   分数差异特征

8. Data Labelling

   资料标示

介绍 (Introduction)

Fractional differentiation (or Fractional derivative or Fractional calculus) is a great idea once you understood all the equations. But the first time when I was reading Chapter 5 of Advances in Financial Machine Learning by Marcos Prado. All the mathematical symbols and equations just push me away from it.

了解所有方程式后，分数微分(或分数微分或分数微积分)是一个好主意。 但是，这是我第一次阅读Marcos Prado撰写的《 金融机器学习进展》第5章。 所有的数学符号和方程式都使我远离它。

I was like, what is this?

我当时想，这是什么？

Believe me, if you feel lost, you are not alone. The reactions of most people just like what Homer Simpson said: “If something’s hard to do, then it’s not worth doing.” Luckily, I was one of the most people until now.

相信我，如果您感到迷路，那么您并不孤单。 大多数人的React就像荷马·辛普森(Homer Simpson)所说：“如果很难做，那就不值得做。” 幸运的是，我是迄今为止最多的人之一。

Let’s face it and it wasn’t as hard as it looks like. We can break it down and figure it out step by step with the first principle.

让我们面对现实吧，它并不像看起来那么难。 我们可以按照第一个原则将其分解并逐步解决。

Ok, let’s break down the whole chapter and talk it through.. together, emmm.…I mean together…

好吧，让我们分解整个章节并进行讨论..一起，emmm ....我的意思是一起...

背景 (Background)

If you have no idea of what derivative is or need to refresh your memory of derivative learned in school. There are a few really great tutorials online.

如果您不知道什么是导数，或者是否需要刷新对在学校学习到的导数的记忆。 在线上有一些非常好的教程。

The paradox of the derivative | Essence of calculus, chapter 2 by 3BLue1brown.

衍生品的悖论| 微积分的本质，第2章，由3BLue1brown撰写。

Fractional Calculus ~ (FC01) An Introduction to Fractional Calculus by Let’s Learn, Nemo!

小数演算〜(FC01) Nemo，让我们来学习小数演算 ！

well, I watched those 2 videos, but I am still having no idea…. what???? well, you can always watch the third one, I guess.

好吧，我看了那两个视频，但我还是不知道…。 什么？？？？ 好吧，我想你总是可以看第三个。

Ok, let’s do it slowly.

好吧，让我们慢慢来。

为什么要进行分数微分？ (Why Fractional Differentiation?)

In previous articles, we have explored the Feature Engineering & Feature Selection and Data Transformation etc.. Why Fractional Differentiation?

在之前的文章中，我们探讨了特征工程和特征选择以及数据转换 等 。 为什么要进行分数微分？

Definition: A fractionally differenced time series (as opposed to integer differencing) retains the memory of the original series (as indicated by the high correlation), but also it is stationary, the mean and variance of those time series should be time-invariant (or not change with time). (e.g. have the same variance, mean, skew, etc).

定义：分数差分时间序列(与整数差分相对)保留了原始序列的记忆(如高相关性所示)，但它是固定的，这些时间序列的均值和方差应该是时间不变的(或不随时间变化)。 (例如，具有相同的方差，均值，偏斜等)。

What is stationary: A stationary time series is one whose properties do not depend on the time at which the series is observed. Thus, time series with trends, or with seasonality, are not stationary — the trend and seasonality will affect the value of the time series at different times.

什么是固定时间：固定时间序列是其属性不依赖于观测时间序列的时间序列。 因此，具有趋势或季节性的时间序列不是固定的-趋势和季节性将影响不同时间的时间序列的值。

Why stationary is important?

为什么平稳很重要？

If the data is not stationary, it would be wrong that all the assumptions we made based on the financial market data are Independent and Identically Distributed (IID). If the data (features) are not “stationary” (in other words, their underlying data generation process changes its characteristics with time) it would be hard to predict expected future date.

如果数据不稳定，那么我们根据金融市场数据所做的所有假设都是独立且完全相同的(IID)，这是错误的。 如果数据(特征)不是“固定的”(换句话说，其基础数据生成过程随时间改变其特征)，则很难预测预期的未来日期。

Sadly, in financial time series data, the volatility of prices can change rapidly from time to time by events, such as on important information releasing day. Both simple percentages return and log returns as well demonstrated in previous articles in these series, remove most of these relationships (low correlation).

令人遗憾的是，在金融时间序列数据中，价格波动可能会随事件(例如重要信息发布日)不时Swift变化。 在本系列的前几篇文章中也演示了简单的百分比回报率和对数回报率，都消除了大多数此类关系(低相关性)。

This leads to an enormous challenge — how can one make the time series stationary while keeping its predictive power (or memory or information for prediction).

这带