For those interested in time series library, we are developing Darts [1], which focuses on making it easy & straightforward to build and use forecasting models. Out of the box it contains traditional models (such as ARIMA) as well as recent deep learning ones (like N-Beats). It also allows to easily train models on multiple time series (potentially scaling to large datasets), as well as on multivariate series (i.e., series made of multiple dimensions). It will soon support probabilistic forecasts as well.<p>[1] <a href="https://github.com/unit8co/darts/" rel="nofollow">https://github.com/unit8co/darts/</a>
<i>The time series feature (TSFeature) extraction module in Kats can produce 65 features with clear statistical definitions, which can be incorporated in most machine learning (ML) models...</i><p>I'd be curious about the performance of these. A time series featurization library I've liked the look of but haven't used for real is catch22: <a href="https://github.com/chlubba/catch22" rel="nofollow">https://github.com/chlubba/catch22</a><p>In particular I like catch22's methodology:<p><i>catch22 is a collection of 22 time-series [that are] are a high-performing subset of the over 7000 features in hctsa. Features were selected based on their classification performance across a collection of 93 real-world time-series classification problems...</i>
What are suggested online courses to learn about multi variable time series forecasting? My skill level is - ok with university level Biometrics but that was 10+ years ago, and I am web/self-taught python for web apps and automating GIS tasks.
Anybody know where the name comes from? I'm hoping this is a tip of the hat at <i>Zero Wing</i>, otherwise you have no chance to survive make your time.
What are some ways to deal with large volumes of variable-length timeseries for real-time predictions? The best solutions I've tried myself all hinge on windowed-feature extraction or LSTMs. It generally works, but starts to fall apart when you're squeezed for data.<p>It seems that almost everywhere you look, every example has just <i>one</i> timeseries that needs to be dealt with. However, since the methods are much more "statistical" in nature, they can actually make meaningful predictions on a single sample.
Sktime looks like a similar but more fully featured package: <a href="https://www.sktime.org/en/latest/" rel="nofollow">https://www.sktime.org/en/latest/</a>
Is Granger causality a common method for the kind of time series analysis made by this library?<p>I worked with this algorithm before so I was curious, but I can't find it in the API.
Kats looks like a useful library, but I’m a bit surprised to see they’re not enabling parallel execution for the numba kernels. Surely FB must have time-series data large-enough they’d see some performance benefits from parallelism in these functions?