Azure Machine Learning Studio: Creating and using API w/Postman and C#
HTML-код
- Опубликовано: 1 ноя 2016
- Dataset: www.ishelp.info/data/bikebuyer...
This playlist (or related videos) is used in two of my online books:
1. Data Analytics and Machine Learning
- Contents: app.myeducator.com/reader/web...
- Purchase: app.myeducator.com/s/1eKHmUCl...
2. Data-Mining Projects and Database Essentials
- Contents: app.myeducator.com/reader/web...
- Purchase: app.myeducator.com/s/2NgsEMdh...
To request an instructor copy, email me at mark.keith@gmail.com
Hi Mark, I usually refrain from commenting. But you have forced me to, this session is amazing and really helpful. Will definitely recommend to a lot of people.
Very good!
Thanks.
Thank you very much Mark ¡¡¡¡¡¡
thank you very much!
Can you please tell me how Your project moved from BuyerBike Prediction to Demo. Why this model is different from the previous video?
This is video was created a long time before I made all of the others. The rest of them were made sort of in sequence. I guess I just never got around to updating this video to make the exact naming and diagram as the others. But you can definitely modify the pills in this video to match what was in the prior and subsequent videos.
Great videos Mark.
Can you please tell me what's the work around for JavaScript? how can I call ML studio API through JavaScript. For Python and C# it's working fine.
Yeah, the CORS issue. I think I found a workaround once. Let me see if I can find it again
hi Mark, thanks for uploading the video's. They are fantastic.
Out of curiousity, can you make the inputs optional? For example in your video if you only entered in the gender? It would make the score more uncertain but it could still predict?
Rory At Work Hmmm...good question. I don’t think so, but I haven’t tried it.
In order to calculate anything, the API would have to use some default value for the other inputs. If you’re missing data for a certain case, you could calculate the average score for each input across the whole population and use that instead. But it would be safer to setup a new model and API that requires only the gender.
That’s because the regression coefficient for gender will be different when other variables are included versus absent.
Does that help?
That does help. We are using averages, but I was convincing myself that the ML should be able to work with fewer inputs but my searches for the answer proved unfruitful. I think I will now go down the route of using ELO instead of averages.
I'm not sure if this would help in your context, but Azure does have a useful tool for predicting missing values. You may already know about this, but the "Clean Missing Data" pill can be used to replace missing values with "Probabilistic PCA" or "MICE". Anyway, good luck!
Thanks for the replies Mark. It's good to know. I am happy with our approach of averaging the last few matches and adding a hint of ELO. We encapsulate the ML API in our own Azure function API and from there get the other required inputs.
nice videos...Do all of these 40 videos necessary for cracking mcsa 70-744? I want to take up test..
Thanks! No, not that I know of. I just made these for a machine learning course I teach to university students. If it’s useful for that exam, then good luck!
What's type of project? I don't find anywhere
So sorry! I need to check comments better. I think I did a basic Windows Forms project. But it could be an empty project or an MVC project or whatever you want
Need to understand the theory behind different models
Savio Azavedo Hi Savio, are you asking me what the theory is for choosing these particular variables predicting PurchaseBike? Or are you just advising that models should be theory-driven?
Compared to what academics are used to, Machine Leaning is quite “a-theoretical.” Or you might call it more of a “grounded theory” approach where we let the data “do the talking” when it comes to theory.
However, that’s not to say that theory should not inform the makeup of a machine learning model. In reality, selecting variables should be a recursive process that begins with the best academic theory available, but gets modified over time as new phenomena emerge and as context demands.
But if you were simply asking what theory I used for these particular variables in this video, the answer is none 😬. Since this is just a demo, I didn’t worry about theory.
Sorry I should have worded my query a bit differently. My doubt was that I need to know the theory behind different prediction models and which one to use when
Like you have used a linear regression model and a decision tree to classify bike purchases. But which models can we use to predict something more specific like stock prices or sales info. A video going through different prediction models and their use would be awesome . Thanks for your response
Hey Savio, so sorry I forgot about this comment! I got a bit busy. Anyway, have you seen my playlist where I go through each predictive model and their assumptions, best uses, etc.? That may be what you're looking for. However, I don't go into a ton of detail, but definitely enough to get you started. The playlist is here: ruclips.net/video/ljHtItiZHho/видео.html
Hi Keith .....Thanks for that .Your tutorials are definitely useful.....Keep making more of them...Cheers
Hi Mark, Do you have any videos about retraining a model in Azure ML Studio ?