SAS enhances data and text mining software

SAS' integrated data mining and text mining capabilities uncover insights quickly from information contained in structured data as well as unstructured data in large document collections.

Technical and business users alike will benefit from new, easy-to-use graphical reports. Power users and novice users will enjoy SAS' streamlined data mart creation, including support for Web mining, Web crawling and blog analysis. SAS data and text mining software enables decision makers to reduce fraud, curb customer attrition and predict risk of a loan default, for example. SAS provides an unmatched breadth of enterprise intelligence offerings spanning analytics and data integration that includes data cleansing, intelligence storage and reporting. This tight integration delivers seamless deployment of predictive models to improve consistency and accuracy of business intelligence.

SAS' collaborative environment allows users to capitalise on data gathered from a greater number of channels: text, Web paths and time series-related transactions. The software then creates unified predictive models that automatically extract the critical information from all of these diverse data sources. Using the tightly integrated SAS Enterprise Intelligence Platform, the mining results can then be plugged into the SAS ETL Server and SAS Enterprise BI Server or other existing data integration or BI reporting systems. More than ever before, SAS makes it easy to deliver valuable business information to users through their preferred channels.

"SAS Enterprise Miner and SAS Text Miner are worth their weight in gold," said Dr. Patricia Cerrito of the University of Louisville. "They've paid for themselves many times over by highlighting cost and quality issues, leading to valuable improvements in healthcare at Louisville hospitals and worldwide."

SAS Enterprise Miner 5.2 and SAS Text Miner 2.3 provide new statistical and visualisation options for uncovering trends and patterns. For example, new interactive graphics enable business analysts to quickly uncover complex relationships. It is now even easier for users to explore mining results using flexible navigation options that search the data for trends and anomalies to gain insights and fresh ideas. New multi-user grid-computing support reduces the time needed to both build and deploy accurate predictive models.

Enhanced SAS Enterprise Miner

Interactive, exploratory data analysis and modification features enable users to easily uncover data quality issues. This creates data that is optimally structured and prepared, enabling users to focus on the model selection process. Interactive data preparation features include:

• Interactive plots that identify outliers and specify valid ranges both for interval and categorical variables.
  
• An interactive expression builder that allows users to incorporate unique custom business rules, including interactions between factors - even complex ones with mathematical representations involving higher-order polynomial terms.
  
• Interactive specification that identifies replacement options for missing or miscoded values, both for input and target categorical variables.
  
• Customised profit-and-loss decision matrices that are integrated into the scoring process to support finer resolutions of decision definitions.

SAS added visualisation options, such as area bar, scatter plot matrix, lattice, parallel axis and 3D charts, to graphically deliver insights. The Tree Desktop Application enhances capabilities for user-defined splits based on any input variable to support data segmentation. Tree branches can be expanded or collapsed interactively. The graphics enable the user to better control tree complexity.

Enhanced Web mining rapidly identifies the most significant Web path sequences to track the navigational patterns of Web site visitors. By including Web traffic behaviour, time series transactions and market basket contents into the modeling process, data miners can build richer prediction models that capture and highlight complex customer behaviours.

SAS Text Miner expands data mining capabilities

In this enhancement, a new macro automatically creates synonym lists from misspellings, including transposed, inserted, and deleted letters, acronyms, and embedded punctuation. A new browser window for terms readily expands synonyms. Singular Value Decomposition, a proven dimensions reduction technique, highlights key words and processes "like terms" together. SAS automates the once tedious set-up process by removing the requirement to manually enter industry specific vocabulary.

Text parsing now spans eight languages: Dutch, English, French, German, Italian, Portuguese, Spanish and Swedish. This flexibility enables a global organisation to gain a 360-degree view of its entire customer base. Language-based parsing is now combined with interactive drill-down and visualisation of concept linking. In this version, any word can be treated as a concept. At the click of a mouse, concepts can be expanded to show associated relationships. This capability, combined with an integrated taxonomy browser, helps deliver deeper insights.

Documents can be analysed in PDF, HTML or any of a dozen other formats. SAS Text Miner can analyse text from call center notes, comments gathered in surveys, Internet information such as blogs and it can perform Web crawling to better support research needs.

With these enhancements, SAS users now have more control to make dynamic adjustments that enhance their mining efficiency. No other data and text mining solution is able to deliver this kind of customisation tailored to specific data. From the automated data mart preparation to the distributed multisystem grid allocation, the entire process has become easier and more scalable.