Model Assisted Statistics and Applications

Increasingly, customers use social media and other Internet-based applications (e.g., review sites) to voice their opinions and to express their sentiments about brands. These reviews influence brand identity, either directly (by affecting consumer behavior) or indirectly (by generating positive or negative word-of-mouth through online social networks). We present an automated methodology that can be used to collect data from popular brand review sites and discussion boards. Customer feedback is then analyzed using best-practices of text mining and supervised sentiment analysis. Strategic implications of customer sentiments are discussed as we explore the role of sentiment analysis on modification of branding strategy.

Consumer IncomeView is the Equifax next-generation income model that estimates a consumer’s individual annual salary/wage income. This model provides an optimal income solution to our clients by incorporating premier multi-source data assets and advanced machine learning modeling techniques. As a result, the new model significantly increases the scorable population rate compared with the predecessor models and significantly improves the prediction accuracy. The results of Consumer IncomeView have been successfully validated by an array of new proprietary accuracy metrics for model performance measurement, both on internal out-of-time and clients’ data. This paper presents the design, development, methodology and main results obtained from the development of the Equifax Consumer IncomeView model.

This paper illustrates a two-stage approach for predicting customer profitability. The first stage is to build a dichotomous model to predict the customer’s likelihood of future purchase. The second stage is to build a model, with continuous target variable, to predict the conditional future profit generated by the customer given he would make a purchase. Both stages involve the utilization of the gradient boosting and neural network data-mining techniques. In each stage, various ensemble combinations are tried and the one resulting in the lowest validation average squared error is chosen to be the stage model winner. The two model winners are subsequently used jointly for the prediction of future profit. In this analysis, Base SAS ® is used for data manipulation and SAS ® Enterprise Miner™ 13.2 is used for predictive modeling. It is evident that this two-stage modeling approach is robust in predicting customer profitability. Managerial and research implications will be highlighted.

Disease classification is a crucial element of biomedical research. Recent studies have demonstrated that machine learning techniques, such as Support Vector Machine (SVM) modeling, produce similar or improved predictive capabilities in comparison to the traditional method of Logistic Regression. In addition, it has been found that social network metrics can provide useful predictive information for disease modeling. In this study, we combine simulated social network metrics with SVM to predict diabetes in a sample of data from the Behavioral Risk Factor Surveillance System (BRFSS). In this dataset, Logistic Regression outperformed SVM with validation data ROC index of 81.8 and 81.7 for models with and without graph metrics, respectively. SVM with a polynomial kernel had validation ROC index of 72.9 and 75.6 for models with and without graph metrics, respectively. Although this did not perform as well as Logistic Regression, the results are consistent with previous studies utilizing SVM to classify diabetes.

Classifier accuracy is extremely important and can be improved by increasing the size of the training data set. However, in experimental studies it might be very costly to survey cases; therefore, limiting sample size to a minimum is essential. Sometimes very large data sets might not contain enough information, and additional computer resources do not improve accuracy. Stopping at the optimal iteration results in the minimum amount of observations being used, possibly saving computational time and sampling costs.

For this reason, a sequential method of training classifiers can be of use. This paper proposes a sequential method that seeks to sample the minimum number of observations necessary to train a classifier to estimate the feasible minimum rate of misclassification, the Bayes error.

Using SAS/IML ® Studio, this method of classifier training proves ideal as it gives the researcher more control over the process by specifying when the sequential procedure should be stopped. It is not restricted to any single method of classification, and it never seeks to obtain an unfeasibly low misclassification rate.

Multicollinearity is a phenomenon in which two or more identified predictor variables in a multiple regression model are co-dependent or highly correlated. The presence of this phenomenon can have a negative impact on the analysis as a whole and can severely limit the conclusions of the research study. This paper reviews and provides examples of the different ways in which multicollinearity can affect a research project, how to detect multicollinearity, and how one can reduce its impact through Ridge Regression.

Data comes in all forms, shapes, sizes and complexities. Stored in files and data sets, SAS ® users know all too well that data can be, and often is, problematic and plagued with a variety of issues. Although today’s statistical software programs are extremely powerful, they are typically not designed to overcome poor quality data. This paper describes and recommends a comprehensive data preparation and fuzzy matching process to follow to enable improved statistical modeling. Statistical techniques are also available for comparing the results of the process.

Most statistical software users are aware that two or more data files can be joined, or combined, without a problem when the data files have identifiers with unique and reliable values. However, many files do not have unique identifiers, or “keys”, and need to be joined using character values, like names or E-mail addresses. To add to the difficulty and confusion, these identifiers might be spelled differently, or use different abbreviation or capitalization protocols. This paper describes a versatile 6-step approach to handling data preparation and fuzzy matching issues for improved statistical modeling. The steps include the identification and understanding of potential matching scenarios; exploring data values and data types; data cleaning and validation; data transformation; traditional merge and join techniques; and an assortment of techniques to successfully merge, join and match less than perfect, or “messy”, data by doing phonetic matching using special-purpose character-handling functions like the SOUNDEX algorithm, and the SPEDIS, COMPLEV, and COMPGED fuzzy matching functions. Although the programming techniques described in this paper are illustrated using SAS code, many, if not most, of the techniques can be applied to any software platform that supports character-handling capabilities.

Emergency department (ED) overcrowding is a problem for EDs world-wide, but first and foremost it is a problem for the patients who are affected from it as ED overcrowding leads to higher morbidity and mortality for the patients. Forecasting of ED inflow for planning ED operations and steering patients streams in a region is an integral part of mitigating the overcrowding issue. In the study at hand, four different forecasting models for ED inflow are developed, validated and tested. Gradient boosting as modelling method yielded the best results for forecasting the coming 72 hours of ED inflow, mean absolute percentage error of 21%. This is a sufficiently accurate model to implement in a new information system for planning ED operations in a health care region.