Azure Machine Learning vs. Spotfire vs. TensorFlow

For [a] data scientist require[d] to build a machine learning model, so he/she didn't worry about infrastructure to maintain it.

All kind of feature[s] such as train, build, deploy and monitor the machine learning model available in a single suite.

If someone has [their] own environment for ML studio, so there [it would] not [be] useful for them.

A high level of data integration is available here it supports various data sources and so on. Collaborating features allow users to give access to the dashboard and merge data analytics with other team members. It can meet the demands of both small and large size business enterprises. A customized dashboard and reports are provided to meet the specific needs and get support of extensibility through APIs and customized scripts.

Incentivized

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).

Incentivized

• User friendliness: This is by far the most user friendly tool I've seen in analytics. You don't need to know how to code at all! Just create a few blocks, connect a few lines and you are capable of running a boosted decision tree with a very high R squared!
• Speed: Azure ML is a cloud based tool, so processing is not made with your computer, making the reliability and speed top notch!
• Cost: If you don't know how to code, this is by far the cheapest machine learning tool out there. I believe it costs less than $15/month. If you know how to code, then R is free.
• Connectivity: It is super easy to embed R or Python codes on Azure ML. So if you want to do more advanced stuff, or use a model that is not yet available on Azure ML, you can simply paste the code on R or Python there!
• Microsoft environment: Many many companies rely on the Microsoft suite. And Azure ML connects perfectly with Excel, CSV and Access files.

Incentivized

• It has the best coding integration (python, R) of any BI product
• The ability to work with very large datasets (10 mil+) is better than competitors
• Export options are more complete and have better functionality
• The data canvas is the best tool to join and transform data vs. competitors

Incentivized

• A vast library of functions for all kinds of tasks - Text, Images, Tabular, Video etc.
• Amazing community helps developers obtain knowledge faster and get unblocked in this active development space.
• Integration of high-level libraries like Keras and Estimators make it really simple for a beginner to get started with neural network based models.

Incentivized

• It would be great to have text tips that could ease new users to the platform, especially if an error shows up
• Scenario-based documentation
• Pre-processing of modules that had been previously run. Sometimes they need to be re-run for no apparent reason

Incentivized

• The donut chart is I guess a powerful illustrations but I hope it should be done quite simple in Spotfire. But in Spotfire there are lots of steps involve just to build a simple donut chart.
• Table calculation (like Row or Column Differences) should be made simple or there should be drag and drop function for Table Calculation. No need for scripting.
• Information Link should be changed. If new columns are added to the table just refreshing the data should be able to capture the new column. No need extra step to add column

Incentivized

• RNNs are still a bit lacking, compared to Theano.
• Cannot handle sequence inputs
• 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.

-Easy to distribute information throughout the enterprise using the webplayer. -Ad hoc analysis is possible throughout the enterprise using business author in the webplayer or the thick client. -Low level of support needed by IT team. Access interfaces with LDAP and numerous other authentication methods. -Possible to continually extend the platform with JavaScript, R scripts, HTML, and custom extensions. -Ability to standardize data logic through pre-built queries in the Information Designer. Everyone in the enterprise is using the same logic -Tagging and bookmarking data allows for quick sharing of insights. -Integration with numerous data sources... flat files, data bases, big data, images, etc. -Much improved mapping capability. Also includes the ability to apply data points over any image.

Easy and fastest way to develop, test, deploy and monitor the machine learning model.

- Easy to load the data set

-Drag and drop the process of the Machine learning life cycle.

Basic tasks like generating meaningful information from large sets of raw data are very easy. The next step of linking to multiple live data sources and linking those tables and performing on the fly analysis of the imported data is understandably more difficult.

Incentivized

Support of multiple components and ease of development.

Incentivized

Even though, it's a rather stable and predictable tool that's also fast, it does have some bugs and inconsistencies that shut down the system. Depending on the details, it could happen as often as 2-3 times a week, especially during the development period.

Generally, the Spotfire client runs with very good performance. There are factors that could affect performance, but normally has to do with loading large analysis files from the library if the database is located some distance away and your global network is not optimal. Once you have your data table(s) loaded in the client application, usually the application is quite good performance-wise.

Support is nonexistent. It's very frustrating to try and find someone to actually talk to. The robot chatbots are just not well trained.

Incentivized

Support has been helpful with issues. Support seems to know their product and its capabilities. It would also seem that they have a good sense of the context of the problem; where we are going with this issue and what we want the end outcome to be.

Incentivized

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.

Incentivized

The instructor was very in depth and provided relevant training to business users on how to create visualizations. They showed us how to alter settings and filter views, and provided resources for future questions. However, the instructor failed to cover data sources, connecting to data, etc. While it was helpful to see how users can use the data to create reports, they failed to properly instruct us on how to get the dataset in to begin with. We are still trying to figure out connections to certain databases (we have multiple different types).

Incentivized

The online training is good, provides a good base of knowledge. The video demonstrations were well-done and easy to follow along. Provided exercises are good as well, but I think there could be more challenging exercises. The training has also gone up in price significantly in the last 3 years (in USD, which hurts us even more in Canada), and I'm not sure it is worth the money it now costs (it is worth how much it cost 3 years ago, but not double that.)

Incentivized

Not sure

The original architecture I created for our implementation had only a particular set of internal business units in mind. Over the years, Spotfire gained in popularity in our company and was being utilized across many more business units. Soon, its usage went beyond what the original architectural implementation could provide. We've since learned about how the product is used by the different teams and are currently in the middle of rolling out a new architecture. I suggest:

• Have clearly defined service level agreements with all the teams that will use Spotfire. Your business intelligence group might only need availability during normal working hours, but your production support group might need 24/7 availability. If these groups share one Spotfire server, maintenance of that server might be a problem.
• Know the different types of data you will be working with. One group might be working with "public" data while another group might work with sensitive data. Design your Library accordingly and with the proper permissions.
• Know the roles of the users of Spotfire. Will there only be a small set of report writers or does everyone have write access to the Library?
• ALWAYS add a timestamp prompt to your reports. You don't want multiple users opening a report that will try and pull down millions of rows of data to their local workstations. Another option, of course, is to just hard code a time range in the backing database view (i.e. where activity_date >= sysdate - 90, etc.), but I'd rather educate/train the user base if possible.
• This probably goes without saying, but if possible, point to a separate reporting database or a logical standby database. You don't want the company pounding on your primaries and take down your order system.

Use of cloud for better execution power is recommended.

Incentivized

It is easier to learn, it has a very cost effective license for use, it has native build and created for Azure cloud services, and that makes it perfect when compared against the alternatives. As a Microsoft tool, it has been built to contain many visual features and improved usability even for non-specialist users.

Incentivized

Spotfire is significantly ahead of both products from an ETL and data ingestion capability. Spotfire also has substantially better visualizations than Power BI, and although the native visualizations aren't as flexible in Tableau, Spotfire enables users to create completely custom javascript visaualizations, which neither Tableau or Power BI has. Tableau and Power BI are likely only superior to Spotfire with respect to embedded analysis on a website.

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

Incentivized

In an enterprise architecture, if Spotfire Advanced Data services(Composite Studio),data marts can be managed optimally and scalability in a data perspective is great. As the web player/consumer is directly proportional to RAM, if the enterprise can handle RAM requirement accomodating fail over mechanisms appropraitely, it is definitely scalable,

Incentivized

• Productivity: Instead of coding and recoding, Azure ML helped my organization to get to meaningful results faster;
• Cost: Azure ML can save hundreds (or even thousands) of dollars for an organization, since the license costs around $15/month per seat.
• Focus on insights and not on statistics: Since running a model is so easy, analysts can focus more on recommendations and insights, rather than statistical details

Incentivized

• It is costly, so not suitable for small scale implementations.
• Dashboards are as good as the developer, so need experience to get most out of it
• You need to be on Spotfire 11 at least to implement out of the box visualizations
• Integration with Python and R is a game changer, it comes very handy to onboard data scientists without much hassle
• performance is exceptionally well.
• Secure

Incentivized

• Learning is s bit difficult takes lot of time.
• Developing or implementing the whole neural network is time consuming with this, as you have to write everything.
• Once you have learned this, it make your job very easy of getting the good result.

Incentivized