Unit8 Talks #8 - Time series forecasting made easy - Introduction to Darts

Unit8 Talks #8 - On technology - Time series forecasting made easy - Introduction to Open-source Darts
Darts is our open source Python library for time series manipulation and forecasting. Among other things, it contains a good collection of forecasting models - from ARIMA to RNNs and convolutional networks, which can all be used through a single API.
In this webinar, we will discuss the reasons why we decided to create Darts, and how it can be used. In particular, we will cover a few examples that will give an overview of the main functionalities, and discuss some of the roadmap for future developments of Darts.
Who should attend?
- Data Scientists
- Forecasting enthusiasts
- Python enthusiasts
Read more on Darts - / darts-time-series-made...
Get in touch [e-mail]
More about Unit8 - unit8.co/
Unit8 on LinkedIn - / unit8.co
Unit8 on Twitter - / unit8co
Unit8 on Instagram - / unit8.co
Unit8 on Medium - / unit8-machine-learning

[@BuraTohNhi]

I am a Sr Consultant to a popular Supply Chain firm and, I am happy to say we are building our ML forecast model in our product using Darts

[@markdregan]

Huge contribution to the community unit8. Great job. Very well executed. 👊

[@Unit8co]

Thank you, Mark, we will keep sharing the good content! By the way, we are inviting you for our forecasting webinar tomorrow czcams.com/video/fWFbFqp8gzY/video.html

[@julien957]

I am in shock. How those young guys make something complicated so easy to understand and in a interesting way. I really enjoyed the video. Thank you!!!

[@fuzzywuzzy318]

thanks for the video and intrudction, really helpful for beginer

[@mahery_ranaivoson]

Thanks Léo and Francesco. I've been looking for this kind of tool for a while.

[@Unit8co]

Thank you Mahery! We're enjoying working on darts! Stayed tuned about the new releases/features - github.com/unit8co/darts

[@harryakb11]

great, thank you !!

[@shantanuhule2689]

great !!! it was very helpful

[@francycharuto]

Great library. Please tell us how to save the trained model for using dart?

[@Unit8co]

Darts - code & tutorials - github.com/unit8co/darts
Darts - intro & article - medium.com/unit8-machine-learning-publication/darts-time-series-made-easy-in-python-5ac2947a8878

[@9989573418]

Great Library. Can you tell me how missing values features work? What is the intuition behind fill_missing_values function? Does it take auto correlation and lags into consideration to impute the missing values in the given time series? Thanks..

[@gauravnarkhede3622]

Its indeed a great work, Thankyou. I want to learn how to implement forecasting model for Multivariate Time series. Can anyone give me reference?

[@liuzhu4163]

Great library! Just one question, I know the trained deep learning models can be saved with "my_model.save_model(path)" command, but is there anyway to save it as h5 or pickle file? The reason for this question is that I was using it on jupyterlab and the session expires every 24 hours, so the "my_model" gets lost when I do my_model.load_model(path) in the new session. Thanks!

[@yashpatil9564]

Can you please show how to do multivariate time series forecasting with exogenous variable using dart

[@luisguillermopardo7792]

Hey. How can we install de library from the console?

[@kolakareem2441]

This is a great achievement to the community. However, is it possible to use Darts for Multivariate multistep LSTM modelling?

[@aniketpande1703]

hi i am looking for the same, please tell if you got anything

[@ibanguniverse811]

Great vid, where can I find Complete Darts tutorial ? Please let me know

[@Unit8co]

Hi, the code and the tutorials can be found on Github -> github.com/unit8co/darts

[@ibanguniverse811]

@@Unit8co Nice tutorial on Github, but I have a question, what is forecast horizon ? Is backtest just saparate data multiple time like cross validation ?

[@ibanguniverse811]

@Léo Tafti thats clearly explained, thanks for your answer
But I still have a question
I am not sure bout the different between covariate support and multivariate support
Let say I have data with 3 predictors (x1, x2, x3) and two targets y1 and y2
in this case, which one is covariate support or multvariate support ?
Or may be one of them refers to some model ?

[@ibanguniverse811]

And how to use Varmax on Darts ?

[@bhaskarareddysannapureddy3741]

Can we use this "DARTS" for multivariate time series forecasting using GNNs?

[@gurmindersinghmalhotra8184]

yes

[@siddheshthakur603]

It would be great if you could improve your documentation and show a real world project, saving weights, processing real world data and just having some basic tutorials to follow will save hours of people's work.