Time Series Forecasting
A modeling task that uses historical time series data to predict future values. Training and validation data must be split in chronological order to avoid data leakage.
Timeline (last 90 days)
MSAR (Modular Scale Autoregressive framework) is a model-agnostic time series forecasting design that forecasts progressively across multiple temporal resolutions, keeping inputs and outputs aligned at each scale.
The time series field has long lacked unified large-scale pretraining corpora, causing models to often be trained for a single dataset or a single domain.
The paper points out that the core challenge faced by DSR and TSF is topological out-of-domain generalization (OODG), i.e., the dynamical regime of the system itself changes, for example transitioning from periodic behavior to chaotic behavior.
Two years of historical data aggregated by quarter yields only 8 data points, and aggregated by month yields only 24 data points.
Time series forecasting can be applied across multiple domains including energy load forecasting, retail demand planning, financial risk monitoring, and industrial predictive maintenance.
时间序列预测不能简单随机划分训练集和测试集,必须按时间顺序切分以避免数据泄露
All Facts (6)
The time series field has long lacked unified large-scale pretraining corpora, causing models to often be trained for a single dataset or a single domain.
70%UnverifiedMSAR (Modular Scale Autoregressive framework) is a model-agnostic time series forecasting design that forecasts progressively across multiple temporal resolutions, keeping inputs and outputs aligned at each scale.
50%UnverifiedThe paper points out that the core challenge faced by DSR and TSF is topological out-of-domain generalization (OODG), i.e., the dynamical regime of the system itself changes, for example transitioning from periodic behavior to chaotic behavior.
50%UnverifiedTwo years of historical data aggregated by quarter yields only 8 data points, and aggregated by month yields only 24 data points.
50%UnverifiedTime series forecasting can be applied across multiple domains including energy load forecasting, retail demand planning, financial risk monitoring, and industrial predictive maintenance.
50%Unverified时间序列预测不能简单随机划分训练集和测试集,必须按时间顺序切分以避免数据泄露
50%