Share your experience using Actian Rush Analytics

I do use IBM SPSS Statistics, and even my students are using it for their projects and reports while working on PhD or Master's degrees. They are analyzing data using it.

In comparison with other software, I found IBM SPSS Statistics to be the best one from my perspective. It has some features, and especially the current version starts to add artificial intelligence techniques and facilitates the analysis. Every new version has new additions for some functions. I use it for whatever analysis is needed, not just a specific one. I use both IBM SPSS Statistics and IBM SPSS Modeler.

Predictive analytics is the most important part of analytics. There are four levels of analytics, starting with descriptive, then diagnostic, then predictive, and after that, there is prescriptive. Predictive is the main core of using statistical packages. Especially now in IBM SPSS Statistics, we start to have neural networks, support vector machines, classification trees, regression techniques, and generalized linear regression. This is the most important predictive analytics. Predictive analytics has four types: the statistical techniques, which is mainly regression; classification trees; neural or artificial intelligence techniques; plus the time series technique.

Syntax is very important, especially as it is now related to the Python language. This is important to use the syntax, especially if you have replication for the process of analysis. The syntax will be important in this case, whether for updating data or for future data. We use the syntax file.

The only function I may need to be added or hope to be added to IBM SPSS Statistics is how to treat unstructured data. This mainly exists with IBM SPSS Modeler, but I do not think it is able to treat something like videos and similar content unless you are using languages like Python inside IBM SPSS Modeler or inside IBM SPSS Statistics. For the menu itself, for the selection, it does not exist. Thinking of the future, I believe that the owners of IBM SPSS Statistics should think about improving the package itself to be able to treat unstructured data.

I started with version 6, so it is now version 30 or more. This was from the 1990s, maybe 1994. That is like 30 years.

In comparison with other software, I found IBM SPSS Statistics to be the best one from my perspective.

There is a package now called JASP that is trying to imitate IBM SPSS Statistics. It has an advantage of being open source or free. This will give competition with IBM SPSS Statistics. I did not try it with huge data. However, I used IBM SPSS Modeler for more than or almost 8 million records. With students now, we can use JASP as it is a free package and it is imitating IBM SPSS Statistics by using the measurement levels: nominal, ordinal, and scale. It is clearer for using this rather than other software, as they are not classifying the measurement level this way.

Data is now transferred from structured data to semi-structured data to unstructured data. IBM SPSS Statistics is mostly working with structured data. However, if you are having unstructured data, IBM SPSS Statistics is not able up to now to work with it. Here we have to use IBM SPSS Modeler as it is able to work with different kinds of data. However, I think for the future, these two packages need to improve or to give the ability to use unstructured data. This is the future of data now. You are now working with social data, such as Facebook and YouTube and similar platforms. This kind of data needs special treatment, which is not included in IBM SPSS Statistics as it uses only structured data.

IBM SPSS Statistics is working well with structured data like regular data from Excel and similar sources. However, when you start to use unstructured data or something such as videos and sound, in this case, you will be in need for a tool that is not able to be used with IBM SPSS Statistics. I would rate this review as an 8.

I have been using IBM SPSS Modeler for a long time.

I am using IBM SPSS Modeler mainly for ETL. Sometimes I use it to compare the results of the modeling as compared to MATLAB. MATLAB is the main tool for predictive modeling as it is much more powerful than IBM SPSS Modeler.

IBM SPSS Modeler is predictive analytics.

IBM SPSS Modeler is a good tool for heuristic purposes. When you want to go into a project and get first results and make the customer happy in a short time, it is a fantastic tool. They have lots of graphic options in IBM SPSS Modeler.

You can use IBM SPSS Modeler for running predictive modeling with good results.

The customer loves it at first glance because it is so easy to handle, and you get a solution pretty soon.

The customer comes to you and says they want to deploy it and make a production out of this, which is very difficult and expensive with IBM SPSS Modeler. With MATLAB, there is no problem. I have a solution, and then I convert my MATLAB solution to C programming language. This I can deploy, and I can check it, and it is MISRA compatible.

It is very easy to deploy it, to go from MATLAB to C or C++, which is actually needed in the car industry. In the car industry, they want to have it in the hardware. You cannot put MATLAB or IBM SPSS Modeler in the hardware of a car, but with C, there is no problem with a microcontroller. They can shoot it into the microcontroller, and I can check it with Polyspace, and it is MISRA compatible, which is an industrial standard. There is nothing similar in IBM SPSS Modeler. I made solutions with IBM SPSS Modeler, and then the customer said they wanted to make a production out of it, and it was not possible.

I stopped with IBM SPSS Modeler 18. It is now 18.6 from what I know at the moment. I do not believe that there is a possibility to design a graphic user interface with it. It is itself a graphic user interface, where you put all sorts of little icons into the display.

I have been using IBM SPSS Modeler for approximately 20 years.

Maybe you can make a production system with IBM SPSS Modeler. But I am talking about deployment into a microcontroller because my customers are from the car industry. They do not want to put a server in a car for €100,000. They have just a microcontroller with a C program, and this is possible with MATLAB.

I had no problems with it. When I used IBM SPSS Modeler in a project with a company as a consultant working with the IT industry, the support was very good.

Positive

I use predictive modeling with SAP too, which is pretty good. It was the same thing with deployment issues. The deployment problem was actually the reason I went over to MATLAB.

Using different tools, I compared SAS with IBM SPSS Modeler. I have compared IBM SPSS Modeler to SAP predictive modeling, and you get pretty much the same results. There is no big difference, but the solution is different, the way of modeling is different. When you use cross-validation as it should be, then you get similar results.

You should put MATLAB into your database. According to Gartner, from what I have seen last year, there was a Gartner four quadrants perspective, and MATLAB was in a very good position.

I remember a very big project for a customer in France. I made a solution; it was a system where we had video data, physiological data, and the point was to predict vigilance by the video data. It was a big MATLAB project, and I finished in one month with deployment.

You can write me an email or call me to discuss MATLAB.

On a scale of 1-10, I rate IBM SPSS Modeler a nine.