SAP Predictive Analytics [EOL] Reviews

We mainly use this program as a customer management system. We're looking at optimizing customer portfolio management and improving the aggregated lifetime values of the portfolio from one period to the next.

Well, this is difficult for me to answer because that's more on the team side. But I think the features of the actual ability to forecast and pull trends and correlations have been really good.

We've looked at the ability of customer churn, propensity to develop customers and ideas of what makes the ideal customer. We are reaching out to try and predict from a database of what customers would be matching.

I've been using SAP Predictive Analytics for only a couple of months now.

I have found that the solution is very stable. 

We haven't looked too much into the scalability of the solution, but we are fairly confident that it would meet whatever scale we're looking at. It wasn't a turn-off for us from the beginning but we haven't actually attempted to scale.

It's possible that my team have used the technical support but I can't say how their service was. I haven't heard any bad reports yet, so I suppose the technical support is good.

No, we haven't used another solution before.

The initial setup was straightforward and we are still busy with the deployment. It is an on-going process and we may need to call in the help of an integrator or consultant to take it to the next level.

I am not sure what the exact costs are, but I believe it is reasonable.

We looked at Alteryx and we ran it for a while but we moved over to SAP Predictive Analytics. Alteryx couldn't do what we wanted it to do. We wanted a template of stuff that had been done previously, and we wanted some ideas on how to get started. We found that Alteryx was very, very weak in that area.

My rating for SAP Predictive Analytics would be an eight out of ten. If I have to be bold, I'll probably say that we're building away hours, and we are actually putting a lot of the actual predicting stuff back into the warehouse. So running it very bi-directionally. So I'm not sure what its integration features are at the moment, but that's an area we're going to look into in the next month or so.

We are a consulting company and we provide services such as those offered by SAP Predictive Analytics. We use this solution to forecast data based on what the customer is interested in predicting. We push all of the data to SAP, perform analysis on it, and generate reports.

At this point, it's too early for me to judge the accuracy and performance of SAP Predictive Analytics because we have only just been working on basic data. Next quarter, I hope we get some more feedback from our customers on larger amounts of work that needs to be done, on the order of millions of records. With that, we will see the accuracy of the models the solution generates. As of now, I'm not able to say because we are still working on it and need more time.

The most valuable features are the analytics and reporting.

This solution works for acquired data but not live, real-time data. If we connect it to a live backend system then we cannot perform predictive analytics on top of that. We have to first upload data to the cloud, manage the staging environment, and then perform the analysis.

In the next release of this solution, I would like to see more automation in generating the models. The system can suggest the dimensions and measures that should be used, and pre-populate some of the information based on that. Power users would not have as much need for this, but this type of automation would be very helpful for business end-users.

We have been using SAP Predictive Analytics for about four months.

I have not experienced any issues.

We always purchase SAP support because it is very good. They give you the maximum attention and it is really quite amazing.

The initial setup is straightforward and simple.

The length of deployment depends on the environment. It is a cloud-based deployment and for us, it was a seamless integration. However, if there are firewalls and other network aspects to consider then it may take four or five days.

A free trial version is available for testing out this solution.

My advice to anybody who is researching this solution is to try the free trial version. I would suggest they explore Discovery Insights and Discovery Reports, as well as how it integrates with other reporting tools. 

This solution has been good for me so far, and I am continuing to explore the analytics.

I would rate this solution a nine out of ten.

1. Ease of use. Can get results quickly.
2. Intuitive understanding of different data types. Saves time and effort.
3. Descriptive statistics of my dataset. Serves as logical justification for the predictions.
4. Presents output in visual form (Predictive Power/Reliability, Key Contributors). Can quickly assess quality of model and gather insights at a glance.
5. Alerts regarding suspicious variables. Enables quick fixes to model.
6. Dataset modeling functionality. Can be used to quickly enrich the dataset using just a few clicks, as opposed to spending a lot of time considering the different enrichment possibilities and then spending the time to build the dataset accordingly.
7. Confusion Matrix. Good for setting the right threshold value.

My organization is a consulting company that helps its clients in using technology to improve their businesses. This product was the first one that we used to demonstrate capabilities to our customers.

1. Dataset modeling functionality could be easier to use.
2. Neural network interface for examining complex data types like pictures, audio, etc. But I suppose the Leonardo – TensorFlow integration might be taking care of that.
3. Data exploration features could be improved, so that instead of only observing the standard deviation or variance of a column, you can also see the skew and kurtosis.
4. Data transformation features. I’m not sure if it is possible to normalize the dataset using log transformations and such, but a quick way to do that would be nice.

Two to three years.

Minor bugs but nothing serious. Performs pretty fast, and doesn’t really crash.

Not applicable.

Have not needed as yet.

We have also used Microsoft Azure Machine Learning.

Both have their pros and cons. I would say SAP Predictive Analytics is better suited for business users because it hides the complexity of the model (parameter tuning, etc.), whereas Microsoft Azure Machine Learning provides a lot more flexibility for technical professionals to tweak the model (e.g. parameter tuning to improve the Precision, Reliability, or the F1 score, which can be useful depending on the business objective).

Very straightforward. But from what I have observed, people not familiar with machine learning concepts don’t get it right away.

The license fee appears to be prohibitively expensive and overly secretive, leading our clients to opt for cloud-based solutions that only charge for data storage and processing time.

Not applicable.

Hire Bilot to do it for you the right way.