时序预测任务

Part1 领域介绍

Time series is a series of data points indexed in time order.

时间序列分析具体包括的任务：

• 检索Indexing (query by content): given a time series and some similarity measure, find the nearest matching time series.
• 聚类Clustering: find groups (clusters) of similar time series.
• 分类Classification: assign a time series to a predefined class.
• 分割Segmentation (Summarization): create an accurate approximation of a time series by reducing its dimensionality while retaining its essential features.
• 预测Forecasting (Prediction): given a time series dataset up to a given time tn, forecast the next values.
• 异常检测Anomaly Detection: find abnormal data points or subsequences.
• 因果分析Rules Discovery: find the rules that may govern associations between sets of time series or subsequences

推荐教材

• Forecasting: Principles and Practice，第三版（英文），第二版（中文）

推荐公开课

• Intel 时间序列分析：讲授时间序列分析，以及用于预测、处理和识别顺序数据的方法。
  • 时间序列和平稳数据简介
  • 数据平滑化、自相关性和自回归积分滑动平均 (ARIMA) 模型等应用
  • 高级时间序列概念，如卡尔曼滤波器 (Kalman Filter) 和傅里叶变换 (Fourier Transformation)
  • 用于时间序列分析的深度学习架构和方法

添加👇微信，拉你进入学习群。

Part2 时序Python库

	Forecasting	Classsification	Anomaly Detection	Segmentation	TSFeature
Prophet	✅
Kats	✅		✅		✅
GluonTS	✅		✅		✅
NeuralProphet	✅		✅		✅
arch	✅
AtsPy	✅
banpei			✅
cesium					✅
darts	✅
PaddleTS	✅				✅

• Kats，推荐指数：⭐⭐
  • 主页：https://facebookresearch.github.io/Kats/
  • Github：https://github.com/facebookresearch/Kats
• darts，推荐指数：⭐⭐
  • 介绍：a Python library for easy manipulation and forecasting of time series. It contains a variety of models, from classics such as ARIMA to deep neural networks.
  • 主页：https://unit8co.github.io/darts/
  • Github：https://github.com/unit8co/darts
• GluonTS，推荐指数：⭐⭐⭐⭐
  • 主页：https://ts.gluon.ai/index.html
  • Github：https://github.com/awslabs/gluon-ts/
• NeuralProphet，推荐指数：⭐⭐⭐⭐
  • 主页：https://neuralprophet.com/
  • Github：https://github.com/ourownstory/neural_prophet

更多的模型介绍可以查阅论文[arxiv 2021]A systematic review of Python packages for time series analysis.

Part3 相关模型

Time Series Forecasting

Model	Univariate	Multivariate	Probabilistic	Multiple-series training
ARIMA	✅		✅
VARIMA	✅	✅
AutoARIMA	✅
ExponentialSmoothing	✅		✅
Theta and FourTheta	✅
Prophet	✅		✅
FFT (Fast Fourier Transform)	✅
RegressionModel (incl RandomForest, LinearRegressionModel and )

Time Series Classification

• LSTM FCN，LSTM Fully Convolutional Networks for Time Series Classification

Anomaly Detection

• [AAAI 2022] Towards a Rigorous Evaluation of Time-series Anomaly Detection

Time Series Representation

• [AAAI 2022] TS2Vec: Towards Universal Representation of Time Series

Data Augmentation

• [IJCAI 2021] Time Series Data Augmentation for Deep Learning: A Survey
• [arxiv 2020] An empirical survey of data augmentation for time series classification with neural networks

Part4 时序数据集

• UCR Time Series Classification Archive
• UEA & UCR Time Series Classification Repository