1. This product is for those who are knowledgeable in the field of analytics.
2. Import/Export functionality using "browse" capability is part of the product.
3. Import/Export functionality for Excel files is already part of this product.
The fact that Google is moving toward Software-as-a-Service in analytics is impressive and the correct direction. Also, Mike points out that this is the first step in a journey to put analytics capabilities in the hands of everyone. The future is here and it is exciting! Enjoy the post, and by all means give this product a try and let me know your opinion (I do not post anonymous opinions).
You can install the engine as an iGoogle gadget by following the link below:http://www.google.com/ig/adde?hl=en&moduleurl=hosting.gmodules.com/ig/gadgets/file/115640297026242314759/adapawidget.xml
I was looking at the new ADAPA iGoogle gadget and this is a tool that would mostly be used by people that have technical knowledge about predictive analytics. There are some basic improvements that must be made so that analytics is available to "everyone" (see my post Business Analytics: Unleashing the Power from Within).
Mike's comments: "Yes, you are correct that ADAPA is a tool for people who are somewhat familiar with predictive analytics. The intention of the ADAPA iGoogle gadget is not to teach "everyone" about predictive analytics, but to deliver a tool that makes it easy to deploy and use such models.
ADAPA bridges the gap between the scientists developing the models, the IT folks who have to integrate & deploy the models, and the business who ultimately want to use the results of the models to drive the business process.
Predictive analytics uses mathematical models that assume a certain technical knowledge and for many users that will always remain a black box. Only in the context of a business application, most users will understand the outcome of a particular model, e.g., that a high risk score means that the related credit application is bad for business.
Following your vision, we should strive towards building many of these iGoogle Gadgets in a form that focuses on the business user and "hides" the technical details.
Our ADAPA iGoogle Gadget was just an example to demonstrate that is actually possible to deliver predictive analytics as service and make it available via simple iGoogle gadgets. We wanted to empower the scientists and engineers to easily build business applications that leverage the potential of predictive analytics."
My wish list for this product:
Keep it simple!
- The Demo is talking about PMML standards, etc. Most users do not care about the technical aspects and standards, they just want to dig in and make it happen. Do not get a technical person to explain how it works, get a sales person to explain how it works. We usually complaint that sales people over simplify technical issues, but this is a case that "sometimes less is better than more". Go back to Google home page and compare it with Yahoo and MSN to understand this concept.
Mike's comments: "In order to understand the basics about predictive analytics, it is critical to know that there is a standard to represent such models. If you want to represent a web page, you need to know that there is something like HTML. If you want to represent a predictive model, you need to know that the common language is PMML and that there is such a file format.
Unfortunately, we still have much work to do in order to evangelize the PMML standard. Many people don't know about this standard which allows models to be shared between various software tools from different vendors."
- The converter should be behind the scenes. Again, just make it happen and do not bore us with the details.
Mike's comments: "We made a conscious decision to keep the PMML converter a separate product, so users can choose to only convert various PMML versions, but not be locked into using our ADAPA deployment platform. It is our support to the PMML community and will remain a free service."
- Explanation for models should be simple. For example: "Linear regression takes numerical values only. Numerical means that if you can add those values and it means something. You can represent months in numbers but if you add those numbers it does not mean anything. Numerical representation of months are not a numerical value."
Mike's comments: "'Explanation' are supposed to explain what the model does, this is an element of the PMML standard (description). Maybe it could be more descriptive, but it is not really the place to explain how a linear regression works."
- I think that the future of analytics is to allow everyone to see the clusters in the data and clarify the driving factors. See, Kaizen and Analytics: The Power of Each Employee to See Data
Google must overcome the basic rule: People would rather live with a problem they cannot solve than accept a solution they cannot understand . I am sure that the science behind the models are good. On the other hand, the science behind this type of algorithms have been out in the market for a long time. The differentiators are: that is free and that it could be used by a company as a Software-as-a-Service (SaaS). See, Software as a Service overview.
Mike's comments: "We are on the same page here. The goal of this application is to make it easy to deploy, integrate, and execute predictive models.
It is the first step towards a more general adoption of predictive analytics. Now we can work towards the same simplicity in business applications that sit on top of the ADAPA deployment platform and draw their decisions from the predictive power of advanced scientific algorithms."