Gluon TS (Time Series)

3 minute read | Last updated 2-Aug-2022 | An icon depicting a retail tag with a heart for 'favourite' AWS Cloud

AWS Labs has a lot of open source code up on GitHub. Gluon TS helps you build and train models for forecasting time series data. It’s built on the Apache MXNet project and is can help with a lot of the heavy lifting for this use case.

I call out a few more details in the Twitter thread below…

Tweet 1/1 Next tweet

let’s look at the Gluon Time Series code repo from @awscloud

it’s up at https://github.com/awslabs/gluon-ts

☁️ #cloud #devops

marknca @marknca tweeted at 16-Nov-2021, 13:06

Start

That thread broke on Twitter for some reason (probably me, I’m probably the reason ). Here’s the rest of it…

Tweet 1/6 Next tweet

this thread is available unrolled at https://t.co/DXLQOJxSHn

yesterday’s threads up at:

Lambda power tools for python, https://markn.ca/2021/aws-lambda-powertools-python/
AWS Deployment Framework, https://markn.ca/2021/aws-deployment-framework/

☁️ #cloud #devops

marknca @marknca tweeted at 16-Nov-2021, 13:06

Tweet 2/6 Next tweet Start

Gluon is build around Apache MXNet, https://mxnet.incubator.apache.org/versions/1.8.0/, which is now an incubator project

original announced in 2017, this one has flown under the radar for most folks…that’s ok, it’s pretty low level but still worth a look

☁️ #cloud #devops

marknca @marknca tweeted at 16-Nov-2021, 13:06

Tweet 3/6 Next tweet Start

this project focuses on “probabilistic time series modelling” which sounds super cool…and is

time series data is everywhere. being able to easily build & train models around this type of data is extremely useful to a lot of businesses

☁️ #cloud #devops

marknca @marknca tweeted at 16-Nov-2021, 13:06

Tweet 4/6 Next tweet Start

we’re talking about output like this

to help you model predictions or actions over time

☁️ #cloud #devops

marknca @marknca tweeted at 16-Nov-2021, 13:06

Tweet 5/6 Next tweet Start

included in the library are some very handy synthetic data generation tools. if you’re testing to ensure that your model is trained on a well rounded data set, these are invaluable

more on them at https://ts.gluon.ai/tutorials/data_manipulation/synthetic_data_generation.html

☁️ #cloud #devops

marknca @marknca tweeted at 16-Nov-2021, 13:06

Tweet 6/6 Next tweet Start

most ppl won’t dive into the level of #ml where you’d need Gluon-TS. the Deeper forecasting algorithm in @awscloud SageMaker usually does the trick: https://docs.aws.amazon.com/sagemaker/latest/dg/deepar.html

…but if you do need it, this is a very handy repo for time series forecasting!

/ ☁️ #cloud #devops

marknca @marknca tweeted at 16-Nov-2021, 13:06

Start