Can you please provide a brief description of your company and the role that you play there?
We’re a small consulting firm specializing in disability claims. We work with insurance companies to model their disability claims in terms of predicting the likelihood of returning to work. I am one of the modeling experts at the company.
What was the business challenge your company was facing that incited the need for this platform?
We predominantly use them for our predictive model; we feed data into Statistica with the goal of predicting the likelihood of recovery. In the insurance world, when you incur a disability claim, whether it’s long-term or short term disability, having a prediction of when that claim may return to work is very useful, especially for the insurance company to manage the claim. We provide Statistica with our historical data of an insurance company’s experience in terms of when their claims return to work. Based on that, we produce predictions on future claims.
How does your company implement this platform or software?
We use Statistica to build the models, then we provide our customers the results of the model. Typically, we perform the modeling and the scoring, so we are the only one that uses Statistica. The clients are our end users in the sense that they see the results, but they don’t know where the results are coming from.
Was your company considering other platforms? Why this platform?
We’ve looked at several software solutions. We have some team members who have experience with SAS [Statistical Analysis Systems], and I think user experience is fairly good. However, the cost is a little bit on the high side for a small company like ours. Another platform that we’ve considered is R, and we do use a little bit of R. On the flip side, R requires a little bit more programming knowledge or a steeper learning curve, making it tougher to work with.
Statistica gives us the best of both worlds. From a user experience, it’s quite easy to use, and it’s easy to train someone new to pick up the software. From a cost perspective, it’s in the middle ground; not too expensive, but not completely free like R is. From our perspective, it gives us the best offer overall. One of the best things that we like about Statistica is that it has a very wide range of modeling tools. Everything from simple regressions, all the way to the more complicated networks and algorithms, are all there.
Can you share any success, metrics, or overall results of your implementation of this platform?
There have been successes, but because I’m not close to the front end. I’m not comfortable quoting any exact metrics in case it’s not quite accurate.
Were there any software features/tools that you were really impressed by?
Our favorite here is the boosting algorithms. We find that, at least for our applications or what we use Statistica for, the boosting algorithms produce superior results relative to the other algorithms. I also like that the work space, which was provided in the last five years or so, has made automation and replication of work a lot easier than it used to be. The workspace type format is now more standard across the different industry platforms, so it’s usually there as opposed to not being there. That’s a feature that we really like. It’s easy to follow someone else’s work.
Looking back, are there any areas of the software that you feel could be added or improved upon?
The lift charts, gains charts, and ROC [receiver operating characteristic] curve, can be generated when you run an algorithm. So if you run an algorithm, those features can be produced as part of the results; however, for instance, if you had predictions that you got from somewhere else or an ensemble of predictions from several other algorithms, then there is no easy way to produce a lift chart, gains chart, or ROC curve from a set of predictions. We had contacted Statistica a while ago about this, but I don’t think they had a plan to put it on.
Have you had to interact with the platform’s support team or reference their support resources?
We haven’t participated in the forums. We’ve worked with some of the Statistica staff, but it’s been a while since we’ve had direct contact. The contact that we did have was very good, but that was before DELL purchased Statistica. Pre-DELL, there seemed to be a lot more newsletters, or things that Statistica itself was publishing. Post-DELL, I haven’t seen much come to us. I used to get some useful tips and information from those newsletters. I also found it easier to call and get something from Statistica pre-DELL. Now it’s a little more cumbersome and less user-friendly.
To sum up: We have a few quick questions, and for each question we ask you to rate the software on a scale of 1-5 with 5 being the best.
What would you give the software for functionality of the features available?
Four and a half.
What would you give the software for ease of use or ease of implementation into your business?
For support, as in responsiveness of the team or helpfulness of the resources available?
How likely are you to recommend the software to a colleague or similar business?
Overall, satisfaction with the platform?