An end-to-end platform for AI, data science, and analytics, used for modeling, as well as management and deployment of AI models.
N/A
TensorFlow
Score 8.9 out of 10
N/A
TensorFlow is an open-source machine learning software library for numerical computation using data flow graphs. It was originally developed by Google.
SAS Advanced Analytics excels with projects that have at least 3 parts. The first part is the ability to address and compare different modeling types. Suppose you are an analyst interested in predicting home prices or whether an individual will reapply for unemployment insurance. There are lots of model types that could work for these two situations. SAS Advanced Analytics makes it easy (although not as easy as SAS Enterprise Miner) to compare the performance of different modeling types, such as comparing support vector machines with random forest models. A second scenario that SAS Advanced Analytics does a good job at is making the analysis reproducible. By showing the lineage of analyses, another analyst is able to follow the work of the previous analyst. This is a huge advantage for individuals working in corporations or governments. The third area SAS Advanced Analytics is useful is in text analytics. The field is huge now, and I haven't come across a software that makes text analytics as easy as SAS Advanced Analytics.
TensorFlow is great for most deep learning purposes. This is especially true in two domains: 1. Computer vision: image classification, object detection and image generation via generative adversarial networks 2. Natural language processing: text classification and generation. The good community support often means that a lot of off-the-shelf models can be used to prove a concept or test an idea quickly. That, and Google's promotion of Colab means that ideas can be shared quite freely. Training, visualizing and debugging models is very easy in TensorFlow, compared to other platforms (especially the good old Caffe days). In terms of productionizing, it's a bit of a mixed bag. In our case, most of our feature building is performed via Apache Spark. This means having to convert Parquet (columnar optimized) files to a TensorFlow friendly format i.e., protobufs. The lack of good JVM bindings mean that our projects end up being a mix of Python and Scala. This makes it hard to reuse some of the tooling and support we wrote in Scala. This is where MXNet shines better (though its Scala API could do with more work).
Complex Survey Analysis- SAS is a great resource if you need to analyze complex survey data. One can easily write code for this by inserting (survey) in front of the procedure with the weight, cluster, and strata variables. (ex: surveyfreq)
Modeling/ Graphing- SAS creates clean and easy to understand graphs and models which take visual data to the next level.
Support- There is a large SAS Advanced analytics online support in place. It is easy to find help on many procedures that you will use in this software.
SAS Analytics does not have very good graphic capabilities. Their advanced graphics packages are expensive, and still not very appealing or intuitive to customize.
SAS Analytics is not as up-to-date when it comes to advanced analytical techniques as R or other open-source analytics packages.
Theano is perhaps a bit faster and eats up less memory than TensorFlow on a given GPU, perhaps due to element-wise ops. Tensorflow wins for multi-GPU and “compilation” time.
Not only does SAS become easier to use as the user gets more familiar with its capabilities, but the customer service is excellent. Any issues with SAS and their technical team is either contacting the user via email, chat, text, WebEx, or phone. They have power users that have years of experience with SAS there to help with any issue.
If SAS Enterprise Guide is utilized any beginning user will be able to shorten the learning curve. This is allow the user a plethora of basic capabilities until they can utilize coding to expand their needs in manipulating and presenting data. SAS is also dedicated to expanding this environment so it is ever growing.
SAS probably has the most market saturation out of all of the analytics software worldwide. They are in every industry and they are knowledgable about every industry. They are always available to take questions, solve issues, and discuss a company's needs. A company that buys SAS software has a dedicated representative that is there for all of their needs.
Although nothing is perfect, SAS is almost there. The software can handle billions of rows of data without a glitch and runs at a quick pace regardless of what the user wants to perform. SAS products are made to handle data so performance is of their utmost important. The software is created to run things as efficiently as SAS software can to maximize performance.
SAS is generally known for good support that's one of the main reasons to justify the cost of having SAS licenses within our organization is knowing that customer support is just a quick phone call away. I've usually had good experiences with the SAS customer support team it's one of the ways in which the company stands out in my view.
Community support for TensorFlow is great. There's a huge community that truly loves the platform and there are many examples of development in TensorFlow. Often, when a new good technique is published, there will be a TensorFlow implementation not long after. This makes it quick to ally the latest techniques from academia straight to production-grade systems. Tooling around TensorFlow is also good. TensorBoard has been such a useful tool, I can't imagine how hard it would be to debug a deep neural network gone wrong without TensorBoard.
SAS has regional and national conferences that are dedicated to expanding users' knowledge of the software and showing them what changes and additions they are making to the software. There are user groups in most of the major cities that also provide multi-day seminars that focus on specific topics for education. If online training isn't the best way for the user, there is ample in-person training available.
There are online videos, live classes, and resource material which makes training very easy to access. However, nothing is circumstantial so applying your training can get tricky if the user is performing complex tasks. When purchasing software, SAS will also allocate education credits so the user(s) can access classes and material online to help expand their knowledge.
Ask as many questions you can before the install to understand the process. Since a third party does the installation your company is sort of a passanger and it is easy to get lost in the process. It also helps to have all users and IT support involved in the install to help increase the knowledge as to how SAS runs and what it needs to perform correctly.
We had major use of SAS in forecasting where it doesn't require high level of coding knowledge and which has highly efficient models built in which can give good results on forecasts without lot of manual intervention. This tool was designed specifically for forecasting and hence was always a better choice compared to other tools.
Keras is built on top of TensorFlow, but it is much simpler to use and more Python style friendly, so if you don't want to focus on too many details or control and not focus on some advanced features, Keras is one of the best options, but as far as if you want to dig into more, for sure TensorFlow is the right choice
It all depends on the type of SAS product the user has. Scaleability differs from product to product, and if the user has SAS Office Analytics the scaleability is quite robust. This software will satisfy the majority of the company's analytic needs for years to come. In addition, if SAS is not meeting the users needs the company can easily find SAS solutions that will.
SAS Advanced Analytics is not the cheapest software on the market. The overall cost was weighed against free, open-source software tools. The overall return, I think, was quite positive because SAS Advanced Analytics saves enormous amounts of time compared to the open-source software tools.
At first, adopting SAS Advanced Analytics was a negative return because it took time for individuals to change their analytics habits and adjust to superior tools available at their discretion.
SAS Advanced Analytics has replaced the need to hire less expensive R or Python programmers. So, although the software requires an initial expensive upfront investment, the ease of use makes it so that other areas of expenditure save money.
