Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Ask HN: Who uses Azure ML?

news.ycombinator.com

Post reply on HN
Print view View original

11–15 of 15 posts

Re: Ask HN: Who uses Azure ML?

#11

Post by a3camero »

BIackSwan wrote: post #9
Avalaxy wrote: post #5

Earlier quoted context omitted.

Very interesting! Bookmarking your blog post for later. I'm still deciding whether I should focus on Scikit/TF/AzureML, or multiples. Any comparisons between them are welcome. Would be nice to see a cost breakdown for each of them either (as in development time and $$ to run it).

Its very contextual. So the answer is "it depends". Azure ML will cost you since they have a non free tier in $$ whereas the other tools can scale without that. BUT for the other tools which are free in $$ you need more powerful hardware available as you start using larger datasets. Also, the ease of use of Azure is a benefit in itself, you don't need to learn any coding library and how to use it - speeds up your ite…

Although you can sign up for Bizspark (or if you have a startup: Bizspark Plus) and use quite a bit for free.

Re: Ask HN: Who uses Azure ML?

#12

Post by conjectures »

Avalaxy wrote: post #3
hcrisp wrote: post #2

To my knowledge, it only works with dataframes, has limited algorithms, and results in non-portable models due to vendor lock-in. That said, it makes ML easier for the non-programmer, is seeing fast growth due to MS investment, and can be well integrated to custom R and Python code.

> and results in non-portable models due to vendor lock-in Yea, I was wondering about this. The free tier ( https://azure.microsoft.com/en-us/pricing/details/machine-le... ) says that publishing it as a production API is not available. I wonder how you can get data out of it without the API. I'm also wondering how the costs in a real life scenario would compare to running your own software on bare metal. Maybe it wou…

Not MS Azure, but AWS - alot of their more advanced APIs seem to rack up bills quite quickly.

Their prediction API is $0.0001 per hit. So if you get r requests per second cost per hour is r0.000160^2. Cost to rent an EC2 server is x per hour.

So if x<r(0.0001)(60^2) it's cheaper to rent. Say you need an r3.large that's x=$0.166. So it's cheaper if you get more than 0.46 requests per second, which sounds like quite a light load.

Re: Ask HN: Who uses Azure ML?

#15

Post by roopalik »

Thanks for your interest in Azure ML! I am part of the product team and I am happy to answer any questions you have.

Specifically on pricing, once the model is built using the Azure ML Studio, the deployed web service can be accessed by using the dev/test tier. The purpose of the dev/test tier is for you to play around with the web service and ensure that it is giving you the results you are expecting. Once you are satisfied with the results and plan to go into production then we have paid tiers starting at 100$ per month (Standard S1) which includes 100K API calls and 25 hours of compute. The exact plan you might need (S1,S2 or S3) depends on your application and the number of calls / compute it uses up. You can change / cancel plans at any time paying only for what you used.

Here [1] is one case study of Track Revenue, a startup using Azure ML to optimize marketing offers for its customer’s campaigns. They are a member of the BizSpark program [2](mentioned earlier on this thread) at MS which offers startups a significant number of free azure credits that can be applied against any Azure service usage including Azure ML.

Let me know if you have any further questions. You can reach me directly at roopalik@microsoft.com.

[1] https://customers.microsoft.com/en-us/story/this-maverick-mo... [2] https://bizspark.microsoft.com/

Post reply on HN