We've all wished that we could peek into the future at one point or another. Whether it's seeing the next upswing in the stock market, finding out how a relationship will work out, or, in this economy, just determining if you'll have a job in six months, predicting upcoming events could definitely come in handy. That's why SPSS Clementine data mining software, which predicts future patterns of growth and customer change, is so useful.

Clementine is a data mining workbench that uses predictive techniques to uncover patterns and trends in your data, which can help you improve current processes and make smart decisions for your business. Most organizations have data in multiple locations, including databases, spreadsheets, flat files, enterprise resource planning systems, email, and networks. With Clementine, you can collect that information from these seemingly disparate storage options, create models based on the data, and integrate those models into your everyday business in order to reap the benefits of seeing your customer actions before they happen. So if you sometimes feel like you're pinning the tail on the donkey of future business events, Clementine might be the solution you've been searching for.

Looking At the Past

With Clementine, companies can fully interact with customers at every level in order to determine past and future behaviors. The solution helps you utilize all your data assets to create a representative view of your customers or constituents. Clementine works by collecting the chosen data (including demographic information such as age, gender, or region, as well as purchase behaviors), routing it through modeling algorithms, and using the results to create business predictions or just to identify key customer segments.

The solution can help you make predictions involving future growth and potential risks. For example, you can provide predictive results to salespeople so they can offer products that customers are most likely to purchase. You can also alter your website in order to feature the products most likely to gain customer attention and site hits.

If you have data you feel is important but that isn't included in your data stores, you can add the info into the tool by hand. Just use Clementine's structured format to describe relationships between concepts, attitudes, people, organizations, and events. Then include this information in your models in to order to produce the most relevant predictions. With Text Mining for Clementine, you can even incorporate free-text data from documents, emails, blogs, RSS feeds, and other text sources in your analyses, so nothing slips through the cracks. The new Web Mining for Clementine option even lets you add information about website behavior to your analyses.

To help you feel at ease with putting all your data into SPSS's hands, Clementine provides explicit support for the Cross-Industry Standard Process for Data Mining (CRISPDM), the industry-standard methodology that ensures timely, reliable results with data mining. Clementine 12.0 also integrates with SPSS Predictive Enterprise Services, an enterprise-level predictive platform that lets organizations manage analytical results, automate analytical processes, and deliver predictions. Additionally, Clementine integrates with SPSS Dimensions data collection software, making it simple for companies to access and use attitudinal data from customers, too.

Interpreting Your Data

So now that you've collected all this data, how do you view and organize it in a meaningful way? With the solution's graphical user interface, analysts can spend less time interpreting data and instead use the product's interactive model creator to produce a visual graph or chart of the predictive data. They can stop the modeling process at any point to alter the stream and ensure that models are appropriate and useful.

Clementine also offers visualization capabilities to guide you to more rapid results. Imagine presenting your boss with a jumbled slew of numbers on a page and explaining your insights and requiring an emergency investment based on those numbers. It might be a hard sell. But with Clementine, you can display both the collected and predictive data in easy-to-understand charts and graphs that include everything from histograms, distributions, and line plots to advanced graphs with automatic assistance (via the solution's Graphboard node feature). Users can explore patterns in data by creating new variables from information on an existing graph. The solution's visualization engine even lets you change the appearance of your graphs after you create them.

You can also import models designed with Predictive Model Markup Language (PMML) from tools such as AnswerTree and SPSS Statistics; evaluate models—singly or in groups—using lift, gains, profit, response, and other model evaluation graphs; and use propensity scores for consistent deployment and comparison of disparate model types.

Clementine lets data miners, analysts, and other business users collaborate on projects by combining multiple predictions, because you can use the output from one prediction as the input for a new model. For example, if one model predicts decreased sales in a certain sector, you can use that prediction to create a model that will predict the impact the decrease will have on web traffic advertisement. By seeing how seemingly unrelated aspects of your business affect each other, you can determine which business decisions are going to have big impacts.

Clementine offers a number of algorithms for clustering, classification, association, and prediction, as well as algorithms for automated multiple modeling, time-series forecasting, and interactive rule building. These algorithms exist in a Clementine “base” module and optional additional modules. Whichever modules you choose, you can build models quickly with Clementine. Then you can test the models, explore their behavior, and find the best-performing ones.

Putting Your Predictions to Work

Clementine’s open architecture doesn't just let your organization make future predictions—it can also help your organization analyze many variables related to risk. You can use techniques including risk scoring and anomaly detection to locate suspicious circumstances. For example, one tax agency developed predictive scores with Clementine that helped it to focus on the organizations most likely to have underpaid their taxes, ultimately leading to devoting investigative resources to those cases and increasing revenue without increasing staff.

Erin Bradford is a freelance editor who specializes in IT.