JMIR PUBLIC HEALTH AND SURVEILLANCE Hoyt et al

Original Paper

IBM Watson Analytics: Automating Visualization, Descriptive, and

Predictive Statistics

Robert Eugene Hoyt1*, MD, FACP; Dallas Snider2*, PhD; Carla Thompson3*, EdD; Sarita Mantravadi1*, MS, MPH,
CPH, CHES, PhD
1
College of Health, Department of Health Sciences and Administration, University of West Florida, Pensacola, FL, United States
2
College of Science and Engineering, Department of Computer Science, University of West Florida, Pensacola, FL, United States
3
College of Education and Professional Studies, Community Outreach Research and Learning (CORAL) Center, University of West Florida, Pensacola,
FL, United States
*
all authors contributed equally

Corresponding Author:
Robert Eugene Hoyt, MD, FACP
College of Health
Department of Health Sciences and Administration
University of West Florida
11000 University Parkway
Pensacola, FL, 32514
United States
Phone: 1 8503845235
Fax: 1 850 474 2173
Email: [email protected]

Abstract
Background: We live in an era of explosive data generation that will continue to grow and involve all industries. One of the
results of this explosion is the need for newer and more efficient data analytics procedures. Traditionally, data analytics required
a substantial background in statistics and computer science. In 2015, International Business Machines Corporation (IBM) released
the IBM Watson Analytics (IBMWA) software that delivered advanced statistical procedures based on the Statistical Package
for the Social Sciences (SPSS). The latest entry of Watson Analytics into the field of analytical software products provides users
with enhanced functions that are not available in many existing programs. For example, Watson Analytics automatically analyzes
datasets, examines data quality, and determines the optimal statistical approach. Users can request exploratory, predictive, and
visual analytics. Using natural language processing (NLP), users are able to submit additional questions for analyses in a quick
response format. This analytical package is available free to academic institutions (faculty and students) that plan to use the tools
for noncommercial purposes.
Objective: To report the features of IBMWA and discuss how this software subjectively and objectively compares to other data
mining programs.
Methods: The salient features of the IBMWA program were examined and compared with other common analytical platforms,
using validated health datasets.
Results: Using a validated dataset, IBMWA delivered similar predictions compared with several commercial and open source
data mining software applications. The visual analytics generated by IBMWA were similar to results from programs such as
Microsoft Excel and Tableau Software. In addition, assistance with data preprocessing and data exploration was an inherent
component of the IBMWA application. Sensitivity and specificity were not included in the IBMWA predictive analytics results,
nor were odds ratios, confidence intervals, or a confusion matrix.
Conclusions: IBMWA is a new alternative for data analytics software that automates descriptive, predictive, and visual analytics.
This program is very user-friendly but requires data preprocessing, statistical conceptual understanding, and domain expertise.

(JMIR Public Health Surveill 2016;2(2):e157) doi: 10.2196/publichealth.5810

KEYWORDS
data analysis; data mining; machine learning; statistical data analysis; natural language processing

http://publichealth.jmir.org/2016/2/e157/ JMIR Public Health Surveill 2016 | vol. 2 | iss. 2 | e157 | p. 1

(page number not for citation purposes)
XSL• FO
RenderX
 JMIR PUBLIC HEALTH AND SURVEILLANCE Hoyt et al

New features are added frequently. IBMWA has 4 basic

Introduction sections: refine, explore, predict, and assemble that are described
Studies have shown that physicians tend to lack data analytical in the following paragraphs [7].
expertise, most likely due to insufficient training in statistics Refine
while in medical school or not using statistics on a regular basis
The refine section is used for data exploration and manipulation.
[1,2]. Gigerenzer et al conclude that there is “collective
This is a logical starting point to examine any dataset. Here,
statistical illiteracy” in health care today [3]. Moreover, there
spreadsheet rows and columns are displayed. Attributes can be
is evidence that teaching statistics to students outside of their
renamed, calculations can be embedded, and data can be placed
health care specialty area has been relatively unsuccessful [4].
in groups or hierarchies for subgroup analysis. Attributes can
Contrastingly, the need for data analytical expertise by health
be organized into ascending or descending order, and a data
care personnel has grown exponentially, given the explosion of
score and percent missing data per column is displayed.
health care data in the past decade [5]. The situation is
complicated by a lack of trained data scientists to analyze new Explore
data. According to McKinsey Global Institute, there will be a The explore section is used for descriptive analytics and is
shortage of 140,000 to 180,000 data scientists by 2018 [6]. demonstrated in the next section. Using the natural language
Therefore, new tools are needed to assist health care workers processing (NLP) function of Watson Analytics, a user can enter
in analyzing health data. Clinicians and other health care other questions in the search window. In addition to the map
workers would benefit from tools that could produce descriptive, view, data can be represented in tree, heat, grid, area, bar,
predictive, and visual analytics more rapidly and easily than bubble, line, pie, and categorical charts. IBMWA recommends
tools currently available in most analytical software packages the optimal display type. The page can be saved for a dashboard,
and with little training required for users. or shared via email, social media, or downloaded. In addition,
a hyperlink can be created for the page for remote viewing.
One new tool with potential benefit to health care workers is
IBM Watson Analytics (IBMWA), introduced in 2015. Unlike Predict
the Watson that won Jeopardy in 2011, this platform is not The predict section is used for predictive analytics. The user
based on cognitive computing; instead, it is based on a selects the target attribute and IBMWA generates a predictive
practitioner approach to using advanced statistics [7]. IBMWA strength. For categorical targets, the predictive strength is the
is available in free, personal, and professional versions. An proportion of correct classifications, and for continuous targets
IBMWA Academic Program was also released in 2015 that it is 1-relative error. Data quality is rated at the top of the page
provides free access to the professional version of the program with mention of any outliers, skewed distributions, and missing
for faculty and students, if used for noncommercial purposes data. A user can request an analysis with a single factor, 2
[8]. factors, or all factors. The predict results can be saved and
shared, similar to the explore function. Hyperlinked statistical
In this paper, we will report on the features of IBMWA and
details are available that provide the statistical test used, the
discuss how this software subjectively and objectively compares
statistical significance, and effect size.
with other data mining programs.
Assemble
Methods The assemble section allows for data visualization and
dashboard creation. This function creates dashboards,
Software Packages infographics and slide shows by simply dragging and dropping
IBM Watson Analytics data into the active panes. Multiple choices exist for users to
have options for representing or displaying data [7].
IBMWA utilizes advanced statistics based on Statistical Package
for the Social Sciences (IBM SPSS, IBM Corporation), and the IBM Statistical Package for the Social Sciences
statistical tests used are enumerated in Table 1 [7,9]. The SPSS is a comprehensive statistical package available in
available tests were chosen to provide analysis of nonparametric standard, professional, and premium versions. Table 3 outlines
(nominal and ordinal) and parametric (interval and ratio) types the features of the professional version [9]. The basic or standard
of data. Statistical significance is presented with standard P SPSS software package provides a user-friendly, menu-driven
values in addition to effect size. approach for analyzing data without the need for programming
The features available in the professional (and academic) skills or formula-driven guidelines, options for applying multiple
versions are summarized in Table 2. IBMWA also provides analyses, and generating graphs and visuals to meet the needs
access to additional data sources inherent in social media (such of social science researchers with minimal statistical preparation
as Twitter feeds) and an enhanced ability to share datasets. skills. Unlike Microsoft Excel, SPSS provides a direct use and
understanding of variables and analyses relative to research
principles and interpretation options rather than a formula-driven
and spreadsheet (location-dependent) approach to data analyses.

http://publichealth.jmir.org/2016/2/e157/ JMIR Public Health Surveill 2016 | vol. 2 | iss. 2 | e157 | p. 2

(page number not for citation purposes)
XSL• FO
RenderX
 JMIR PUBLIC HEALTH AND SURVEILLANCE Hoyt et al

Table 1. IBM Watson statistical tests.

Statistical test Indication
Analysis of variance (ANOVA) ANOVA tests mean differences among 2 or more groups and whether the mean target value varies across
combinations of categories of 2 inputs; If the variation is significant, there is an interaction effect
Asymmetry index Ratio of skewness to the standard error
Chi-square automatic interaction detec- Decision tree using chi-square for prediction
tor classification tree
Chi-square automatic interaction detec- Decision tree using chi-square and regression for prediction
tor regression tree
Chi-square tests Using chi-square to compare frequencies in groups, independence, and marginal distributions
D’Agostino’s K-squared test of normal- Determines if normal distribution is present
ity
Distribution test Chi-square test compares conditional distributions with overall distribution
Fisher r-to-t test Transforms Pearson’s r to t test for significance
High low analysis Partitions categories into high or low groups for analysis
Influence test Chi-square test determines whether the number of records in a group is significantly different from the ex-
pected frequency.
Model comparison test Tests whether the key driver has an effect on the logistic regression
Paired samples t test Dependent t test checks whether the means of 2 continuous fields are statistically different or if there is a
change in means over time for one group
Unusually high or low analysis Determines which categories or combinations of categories across categorical fields have unusually high or
low target mean values

Table 2. Features of IBM Watson Analytics Professional.

Features IBM Watson Analytics Professional
Maximum number of rows per dataset 10,000,000
Maximum number of columns per 500
dataset
Input in .csv, .xls or, .xlsx formats Uploaded from PC, Dropbox, IBM Cognos, Box, and Microsoft OneDrive
Data connections IBM Cognos BI server, IBM dash DB, IBM DB2, IBM SQL, Microsoft SQL server, MySQL, Oracle, and
PostgreSQL
Storage 100 GB; can be increased in increments of 50 GB

http://publichealth.jmir.org/2016/2/e157/ JMIR Public Health Surveill 2016 | vol. 2 | iss. 2 | e157 | p. 3

(page number not for citation purposes)
XSL• FO
RenderX
 JMIR PUBLIC HEALTH AND SURVEILLANCE Hoyt et al

Table 3. Features of SPSS.

Tool Function
Core stats and graphics Standard statistical tests for nominal, ordinal, interval, and ratio data
Integration with R and Python lan- Expands programmability involving additional languages
guages
Multiple linear and mixed modeling Analyze complex relationships
Nonlinear regression Predictions on nonlinear data
Simulation modeling Build risk models when inputs are uncertain
Geospatial analytics Integrate and analyze time and location data
Customized tables Analyze and report on numerical and categorical data
Charts, graphs, and mapping Assist reporting capabilities
Missing value analysis Address missing data, imputation, etc
Advanced data preparation Identify data anomalies
Decision trees Identify group relationships to predict future events
Forecasting techniques Predict trends with time-series data
SPSS Text Analytics This add-on complementary software package accompanies SPSS to provide qualitative data analyses and
visuals for quantitative data simultaneously analyzed with SPSS

Table 4. Features of Microsoft Excel Analysis ToolPak.

Tool Function
Analysis of variance (ANOVA) Determines variance on single or multiple factors and mean differences among 2 or more groups
Correlation Determines if a pair of variables are related
Covariance Determines if a pair of variables move together and mean differences in 2 or more groups when controlling
for initial group differences
Descriptive statistics Determines central tendency and variability in the data
Exponential smoothing Predicts a value based on prior forecast
F test for 2-sample variances Performs a 2-sample F test to compare population variances and mean differences relative to variability
testing
Fourier analysis Transforms time-based patterns into cyclical components
Histogram Calculates frequencies of values in dataset
Moving average Forecasts values based on prior averages
Random number generation Fills a range with independent random numbers
Rank and percentile Creates a table with ordinal and percentile ranks and used with chi-square analyses
Regression Linear regression based on “least squares” method
Sampling Creates a sample from a population
t test Tests for equality of population means, with equal and unequal variances based on 1-group or 2-group datasets
z test Performs a 1-sample z test for population comparison or a 2-sample z test for means with known variances

Table 5. Features of Microsoft SQL Server Analysis Services.

Tool Function
Multiple data inputs Use tabular data, spreadsheets, and text files
Data management Data cleaning; management; and extract, transform, and load
Model testing Use cross-validation, lift, and scatter charts
Data mining algorithms Clustering, Naïve Bayes, decision trees, neural networks, regression, and association rules
Scripting language support Mining objects are programmable

http://publichealth.jmir.org/2016/2/e157/ JMIR Public Health Surveill 2016 | vol. 2 | iss. 2 | e157 | p. 4

(page number not for citation purposes)
XSL• FO
RenderX
 JMIR PUBLIC HEALTH AND SURVEILLANCE Hoyt et al

Table 6. Features of Waikato Environment for Knowledge Analysis.

Tool Function
Preprocess Descriptive statistics and ability to preprocess data; Data from .csv and .arff files, web data, database data,
and ability to generate artificial data
Classify Classify data from Bayes, neural networks, regression, decision trees, production rules, and other algorithms
Cluster 12 clustering algorithms, to include the common simple k-means
Associate Association rules for pattern recognition in data
Select attributes Searches for best set of attributes in dataset
Visualize Visualization of data into graphs, etc

County Health Rankings for Florida

Microsoft Excel Analysis ToolPak
The dataset used to demonstrate IBMWA features was derived
The ToolPak is a spreadsheet add-on that provides the features
from the publicly available 2014 County Health Rankings for
found in Table 4 [10]. The use of Microsoft Excel Analysis
the state of Florida [13]. The dataset was altered by merging
ToolPak requires the user to acquire and install the add-on as
demographics with health factors and deleting confidence
well as cognitively account for the differences between the
intervals and zscores to reduce the number of attributes
ToolPak spreadsheet-driven program (involving placeholders
(columns). This dataset ranks counties in each state by health
for data locations requiring user-generated formulas to drive
outcomes and health factors, with health factors categorized as
analyses) and SPSS or IBMWA analysis-driven programs,
health behavior, clinical care, social and economic factors, and
whereby data and locations are defined variables with analyses
physical environment. The data matrix was comprised of
formulas inherent in the software programs.
columns consisting of 41 common health factor attributes and
Microsoft SQL Server Analysis Services demographics; the rows consisted of the 67 Florida counties.
Microsoft SQL Server Analysis Services is an integrated Heart Disease Data
platform for data mining that uses relational or cube data in
The dataset used for comparison among the analytical software
multiple formats to provide predictive analytics. A summary of
packages was derived from a well-known and validated
the features is provided in Table 5. Mining choices include
machine-learning repository [14]. The selected dataset focused
clustering, neural networks, decisions trees, and custom plug-in
on the diagnosis of heart disease (presence or absence) based
algorithms. Because Analysis Services is a component in SQL
on 13 common cardiac risk factors or test attributes (columns)
Server’s suite of business intelligence tools, it integrates easily
and 270 instances or rows (patients). The thallium test (cardiac
with the SQL Server database engine and reporting services
scan) attribute had 3 subcategories: 3 = normal test, 6 = fixed
component. Analysis Services provides a confusion matrix, but
defect, and 7 = reversible defect.
requires the user to manually calculate classification accuracy
measures, such as precision and recall. Additionally, Analysis
Services generates a lift chart, in lieu of receiver operating
Results
characteristic curves [11]. Use Cases
Waikato Environment for Knowledge Analysis Explore Option
Waikato Environment for Knowledge Analysis (WEKA) is a The use cases shown in this section are generated using the
free machine-learning software platform developed by the sample data file named 2014 County Health Rankings for the
University of Waikato in New Zealand. This popular program State of Florida. In the explore section, which is used for
is used for data mining utilizing primarily classification and descriptive analytics, Watson Analytics automatically generated
clustering tools consisting of rules, decision trees, and multiple 10 questions based on the data such as “What is the breakdown
other algorithms. WEKA calculates true positive rates, false of % obese by county?” A map of all Florida counties was
positive rates, precision, and recall. WEKA will also create the automatically generated (without user prompting) with % obese
receiver operator characteristic curves and area under the curve noted for each county (Figure 1), as well as the range. The user
[12]. A summary of the features in WEKA is provided in Table can mouse over each county for specific data.
6.
A user can also enter questions in the search window by
Data leveraging the NLP function of Watson Analytics, for example,
The datasets used for comparing IBMWA and prevailing “What is the relationship between % physically inactive and %
software are publicly available datasets. obese by county?” (Figure 2)

http://publichealth.jmir.org/2016/2/e157/ JMIR Public Health Surveill 2016 | vol. 2 | iss. 2 | e157 | p. 5

(page number not for citation purposes)
XSL• FO
RenderX
 JMIR PUBLIC HEALTH AND SURVEILLANCE Hoyt et al

Figure 1. The % obese by Florida county.

Figure 2. The relationship between % physically inactive and % obese by county.

Figure 3. Predictors for factors related to % obese.

http://publichealth.jmir.org/2016/2/e157/ JMIR Public Health Surveill 2016 | vol. 2 | iss. 2 | e157 | p. 6

(page number not for citation purposes)
XSL• FO
RenderX
 JMIR PUBLIC HEALTH AND SURVEILLANCE Hoyt et al

Figure 4. Predict option for nominal category of less than or greater than 30% obese by county.

Figure 5. Dashboard of Florida County Health Rankings.

Assemble Option
Predict Option
The assemble option contains functionality to create dashboards,
The predict option in IBMWA is utilized for predictive analytics.
infographics, and slide shows. An example of an IBMWA
In our analysis of the 2014 County Health Rankings for the state
interactive dashboard display using the dataset, 2014 County
of Florida, 74 associations were noted at the top of the page.
Health Rankings for the state of Florida reflecting the “%
The attribute “children in poverty” was associated with “teen
obesity” by Florida County is depicted in Figure 5. The county
birth rate.” Select “statistical details” and a Pearson correlation
dropdown list is active; such that once a county is selected, the
of .79 with P<.001 and effect size of 0.63 was noted.
percentages change and correspondingly the map changes.
When % obese was selected as the target, predictions were
automatically generated. The top predictor for “% obesity” was
Comparative Study
“% physically inactive” at 69%, but IBMWA recommended The results from the comparative study among the software
the addition of “% African-American,” which increased the packages are presented in the following subsections. The same
predictive ability to 85%. A screenshot of the predict function heart disease dataset was used as the input to each software
results is shown in Figure 3. package and each package provides differing statistics and
measures which are summarized in Table 7.
The “% obesity” column of attributes was then subdivided into
counties with less than or more than 30% obesity reported and IBM Watson Analytics
the predict function was reexecuted. This second analysis used The IBMWA software conducted a logistic regression for
logistic regression and produced household income as the classification purposes. When the target attribute of heart disease
strongest predictor at 88% predictive strength. A chi-square (present or absent) was used in IBMWA, it revealed that the
analysis comparing the categorical variables demonstrated the thallium test had a predictive strength of 76% (percent correct
following: P<.001, effect size = .46 (Figure 4). classification). The thallium test attribute had 3 subcategories:
3 = normal, 6 = fixed defect, and 7 = reversible defect. Based

http://publichealth.jmir.org/2016/2/e157/ JMIR Public Health Surveill 2016 | vol. 2 | iss. 2 | e157 | p. 7

(page number not for citation purposes)
XSL• FO
RenderX
 JMIR PUBLIC HEALTH AND SURVEILLANCE Hoyt et al

on either normal exam or reversible defect on thallium testing, both the intercept only model. However, IBMWA uses the LR
the chi-square test revealed P<.001 and an effect size of 0.48 test to compare models with the reduced fit single predictor
for the normal test and 0.63 for the reversible defect. model as the default setting, whereas SPSS uses the baseline,
intercept only model as the default setting for LR test
When a full model (3 variables) analysis is conducted with
comparison.
logistic regression, the software also conducts a likelihood ratio
test (chi-square) to determine if the addition of the variables A chi-square analysis was also performed using SPSS with a
improves the fit of the model. Predictive strength increases to resulting likelihood ratio of 78% for comparison purposes.
80% (percent correct classification) and statistical significance Based on the normal exam or reversible defect on thallium
of the target predictor of thallium reduced. Interactions between testing, the chi-square test revealed a significant relationship
thallium and the number of vessels calcified on fluoroscopy (χ21= 76.1, P<.001) and an effect size of 0.53 for the normal
were not significant, P<.09. The likelihood ratio test (χ21= 11.08, test and 0.67 for the reversible defect test.
P<.08) was not statistically significant to the 5% significance
level, and thus the reduced model of thallium alone and heart
Microsoft Excel Analysis ToolPak
disease is the best fit. The ToolPak software can only conduct linear regression, not
logistic regression for classification.
IBM Statistical Package for the Social Sciences
Analysis was not performed because a chi-square test would
Binary logistic regression with heart disease as the dependent
have to be manually run between the target attribute and each
variable and thallium as a single predictor was conducted. As
column. The expected values would need to be calculated and
confirmed in the IBMWA results, predictive strength and percent
run against the actual values to arrive at the chi-square result
correctly classified increases as more variables are included in
and P value. This is very labor intensive compared with the
the regression; however, statistical significance reduces.
other platforms tested.
Logistic regression (LR) with 3 predictors—thallium, number
of vessels calcified on fluoroscopy, and the interaction
Microsoft SQL Server Analysis Services
effect—was conducted, illustrating that the predictive strength Data were analyzed using a decision tree and neural network
of the model was 78%, and the interaction effect was not to compare for classification accuracy. To train the classifier
significant. The number of vessels calcified by fluoroscopy and models, 70% of the data was used, whereas the remaining 30%
the thallium test variables were statistically significant with was held out for testing. The decision tree algorithm was chosen
P=.04 and P<.001, respectively. The LR test compared with because of the ease of understanding the results, while a neural
the intercept only model was significant, with χ23=120.5 and network was selected because of the ability to generally produce
better classification results. The decision tree yielded a
P<.001, indicating that the 3 variable model improved model
sensitivity of 0.80 and specificity of 0.78, while neural networks
fit over the intercept only model.
yielded a sensitivity of 0.77 and a specificity of 0.92. Both
Thereafter, forward selection using the LR test was also algorithms have parameters that can be adjusted to improve
conducted for appropriate variable selection, reducing classification accuracy; however, these parameters need to be
collinearity and demonstrating model fit. By the end of the adjusted cautiously to avoid “overfitting” the model.
stepwise forward regression concerning all variables, the LR
test indicated that thallium remained a statistically significant
Waikato Environment for Knowledge Analysis
predictor, as well as gender, type of chest pain, A J48 decision tree was used as the algorithm with 10-fold cross
electrocardiogram results, exercise-related angina, ST wave validation. The outcome was correctly classified 78% of the
depression, and number of vessels calcified by fluoroscopy. time. The precision for the presence of heart disease was 0.931
Percent correctly classified increased to 90%. The variables and recall (sensitivity) was 0.628. Precision for the absence of
gender (χ21=3.9, P=.049), exercise-related angina (χ21=5.7, heart disease was 0.692 and the recall was 0.947.
P=.02), and electrocardiogram results were statistically Summary
significant at the 5% level, whereas, types of chest pain These preliminary informal analyses indicate that the 4
(χ21=13.3, P<.001), ST wave depression (χ21=11.7, P=.001), analytical programs provide similar results using the same
number of vessels calcified by fluoroscopy (χ21=19.9, P<.001), dataset. WEKA does provide a confusion matrix, Kappa statistic,
and receiver operator characteristics curve area statistic, with
and thallium (χ21=15.1, P<.001) were statistically significant
neither of these analytics supplied by IBMWA. WEKA, in
at the 1% level. Comparing this full model with the intercept contrast to IBMWA, includes more than 50 different algorithms,
only model, it was found that χ21= 77.8 and P<.001. These without any recommendations regarding the optimal choice.
results illustrate that the full model had a better model fit than

http://publichealth.jmir.org/2016/2/e157/ JMIR Public Health Surveill 2016 | vol. 2 | iss. 2 | e157 | p. 8

(page number not for citation purposes)
XSL• FO
RenderX
 JMIR PUBLIC HEALTH AND SURVEILLANCE Hoyt et al

Table 7. Results of the comparison of different analytical packages.

Software package Results
IBM Watson Analytics Using logistic regression, the thallium test had a predictive strength of 76% (percent correct classification); chi-
square test revealed P<.001 and an effect size of 0.48 (normal test) and 0.63 (reversible defect)
Statistical Package for the Social Using logistic regression, a full model with thallium, number of vessels calcified on fluoroscopy, and interaction
Sciences test increased the predictive strength to 78%; however, a statistically insignificant chi-square test proved that the
single model using thallium had the better model fit.
χ21= 76.1, P<.001; effect size 0.53 (normal test) and 0.67 (reversible defect)

SQL Server Analysis Services Decision tree analysis yielded a sensitivity of 0.80 and specificity of 0.78, while neural networks yielded a sensi-
tivity of 0.77 and a specificity of 0.92
Waikato Environment for Knowl- Decision tree precision for presence of heart disease was 0.93 and recall (sensitivity) was 0.63; precision for absence
edge Analysis of heart disease was 0.69 and recall was 0.95

data mining using machine-learning techniques, such as neural

Discussion or Bayes networks, and is not appropriate for many big data
Principal Findings sets. The statistical approach is, however, complimentary to the
machine-learning approach. Statistical modeling using a program
According to Dr. Bill Hersh, “Analytics and related activities such as IBMWA usually involves smaller datasets, a hypothesis,
are the future of clinical informatics, realizing the goal of my and a list of assumptions. Machine-learning, on the other hand,
definition of the field, which is the use of information to improve can handle larger datasets, and does not require the same
individual health, health care, public health, and biomedical hypotheses or assumptions. An overview of the existing software
research [15].” To achieve this goal, we will need a well-trained programs supports IBMWA as belonging in the overlap region
workforce and supportive analytical tools. between data mining and statistics as demonstrated in Figure
IBMWA is an analytical program based on SPSS that 6. In addition, this preview of the use of multiple software
automatically generates descriptive, predictive, and visual programs initiates a scholarly conversation concerning the
analytics. This approach is compatible with “greater statistics” considerable overlap between statistics and machine learning
proposed by John Chambers in 1993 [16]. Traditional statistics as exemplified in Figure 6 [17].
can be laborious when manual computation of complicated A comparison of IBMWA with 3 other data analytical software
formulas is required. Steps to simplify this approach will likely resulted in similar, but not identical results. We did not report
be well received. One could argue that automating analytics is results with Microsoft Excel due its inability to perform logistic
a logical progression, similar to using electronic medical regression and the labor intensive nature of the analysis.
calculators, instead of paper calculations or using the “what you
see is what you get” program to create webpages, rather than IBMWA may be a helpful adjunct approach to teaching both
requiring html programming. This program will likely open the statistics and data mining, given its speed, functionality, and
analytics playing field to a larger audience of health care ease of use. Case studies could be presented in the domain of
workers. interest and both the clinical results and statistical methods
could be discussed. The average user would be able to see
The learning curve for IBMWA is much less steep than missing data, skewed distributions, and outliers with minimal
Microsoft Excel Analysis ToolPak, SPSS, WEKA, or Microsoft effort. Large datasets are amenable to data analyses and quick
Server Analysis Services. Instead of needing an extensive response outcomes using IBMWA- a key element in teaching
background in statistics to decide on the statistical method of and learning statistical concepts. In the University of West
choice, this is performed automatically for the user. A busy Florida, IBMWA is used to augment understanding and
health care worker might use this program to gain preliminary applications of statistical concepts in several health informatics
results and then consult an expert in data science or statistics. and computer science courses. The use of NLP to help explore,
IBMWA is able to handle very large datasets and applies the analyze, and visualize datasets would be helpful for most
most common statistical tests required, but does not perform graduate students, regardless of field.

http://publichealth.jmir.org/2016/2/e157/ JMIR Public Health Surveill 2016 | vol. 2 | iss. 2 | e157 | p. 9

(page number not for citation purposes)
XSL• FO
RenderX
 JMIR PUBLIC HEALTH AND SURVEILLANCE Hoyt et al

Figure 6. Overlap between statistics and machine learning.

In addition, IBMWA provides for ease of data exploration,

Limitations creating a conundrum for the average user—with such ease of
For classification purposes with categorical data, IBMWA does data mining and exploration, eventually the user will detect
not offer the user a choice of statistics, but rather selects the chance correlations between variables that appear to be
approach for the user (eg, logistic regression). IBMWA did not significant relationships. IBMWA allows the average user to
aid users in interpreting results, especially the impact of depend on statistical significance as a measure of the strength
unrelated variables and highly correlated variables of the correlation, as well as the correct model; however, a
(multicollinearity). The aim of IBMWA is ease of use but correct correctly specified model might have insignificant, theoretically
interpretations might be eluded due to the lack of odds ratios, relevant predictors. The IBMWA software does not provide
confidence intervals, and an explanation for seemingly statistical significance for each predictor but only the results of
counterintuitive results. the model comparison test (likelihood ratio test).
The combination option in IBMWA makes it easier for users Limitations exist with every analytical platform. Data
to create complex models, evaluating the effect of up to 3 preprocessing (imputation and data quality) is needed and is
variables on the dependent variable. Users may not be able to critical for success. It is estimated that about 80% of the time
determine the best model; for the average user, it becomes harder spent analyzing data, is spent exploring and preparing the data
to determine if the right model is chosen—with the appropriate for analysis [18]. Specifically, data may need cleaning,
theoretically related variables. For example, with multiple integration, reduction, or transformation before analysis
predictors in a classification analysis, the chi-square value procedures [19]. In addition, data must be of sufficient quality
provided indicates if model fit improves with complexity and and volume to answer a question. Finally, domain expertise and
increased variables (likelihood ratio test result). Predictive appropriately prepared researchers are needed to ask appropriate
strength increases as additional variables are added; however, questions and interpret results, regardless of the analytical
this is counterintuitive and easily confused as decreased platform used.
statistical significance (low chi-square value) with increased
predictors—leading to inaccurate interpretation. In IBMWA, Conclusions
as variables are added and complexity increases, the model IBMWA is a new and interesting analytical tool that may be of
starts to become confusing to interpret without taking the extra value to multiple types of health care workers; however, no
step of checking the fine print definitions; the chi-square and statistical program will replace the time needed for
P- value provided in a full model is actually for comparing preprocessing data and asking pertinent questions regarding the
goodness of fit of a statistical model (LR test), not a measure dataset, but the time spent on analytical processes will be greatly
of variable statistical significance. Users may initially focus on expedited. The IBMWA approach needs to be compared and
the fact that predictive strength will increase each time another contrasted with other approaches and by a diverse group of users
variable is added to the model, resulting in a model that is overly to better understand its role within the analytics realm. Clearly,
complex with unrelated variables. For early health data analysts, IBMWA has limitations but IBM is making frequent changes
the issue of highly correlated variables (multicollinearity) and to this program so users can expect more functionality in the
model fit is hard to detect and interpret. Too many variables future. Additionally, IBMWA may motivate educators and
can result in predicting noise, rather than the dependent variable practitioners to question if it is potentially effective as an adjunct
(overfitting). The LR test acts as a warning sign for users—don’t in teaching statistics and analytics in health care.
add more variables in the model just to increase predictive
strength.

Conflicts of Interest
None declared.

http://publichealth.jmir.org/2016/2/e157/ JMIR Public Health Surveill 2016 | vol. 2 | iss. 2 | e157 | p. 10

(page number not for citation purposes)
XSL• FO
RenderX
 JMIR PUBLIC HEALTH AND SURVEILLANCE Hoyt et al

References
1. Wegwarth O, Schwartz LM, Woloshin S, Gaissmaier W, Gigerenzer G. Do physicians understand cancer screening statistics?
A national survey of primary care physicians in the United States. Ann Intern Med 2012 Mar 6;156(5):340-349 [FREE Full
text] [doi: 10.7326/0003-4819-156-5-201203060-00005] [Medline: 22393129]
2. Manrai AK, Bhatia G, Strymish J, Kohane IS, Jain SH. Medicine's uncomfortable relationship with math: calculating
positive predictive value. JAMA Intern Med 2014 Jun;174(6):991-993 [FREE Full text] [doi:
10.1001/jamainternmed.2014.1059] [Medline: 24756486]
3. Gigerenzer G, Gaissmaier W, Kurz-Milcke E, Schwartz LM, Woloshin S. Helping doctors and patients make sense of
health statistics. Psychol Sci Public Interest 2007 Nov;8(2):53-96 [FREE Full text] [doi: 10.1111/j.1539-6053.2008.00033.x]
[Medline: 26161749]
4. Yimaz M. Amstat. 1996. The Challenge of Teaching Statistics to Non-Specialists URL: http://www.amstat.org/publications/
jse/v4n1/yilmaz.html [accessed 2016-01-20] [WebCite Cache ID 6f3StcC1m]
5. DeGaspari J. Managing the Data Explosion. Healthcare-informatics. 2013. URL: http://www.healthcare-informatics.com/
article/managing-data-explosion [accessed 2016-02-04] [WebCite Cache ID 6f3UHVtLq]
6. Davis P. McKinsey Report Highlights the Impending Data Scientist Shortage Pivotal Blog. 2013 Jul 23. URL: https://blog.
pivotal.io/data-science-pivotal/news/mckinsey-report-highlights-the-impending-data-scientist-shortage [accessed 2016-02-04]
[WebCite Cache ID 6f3TQiGxE]
7. IBM. IBM Watson Analytics URL: http://www.ibm.com/analytics/watson-analytics/ [accessed 2016-05-20] [WebCite
Cache ID 6f3URVTtm]
8. IBM. IBM Watson Analytics Academic Program-WAP URL: https://watson.analytics.ibmcloud.com/solutions/industry/
education/wap [accessed 2016-06-15] [WebCite Cache ID 6f3V7K6s2]
9. IBM. IBM SPSS Software URL: http://www-01.ibm.com/software/analytics/spss [accessed 2016-01-02] [WebCite Cache
ID 6f3XEEkOU]
10. Use the Analytics ToolPak to Perform Complex Data Analysis. Microsoft; 2016. URL: https://support.office.com/en-us/
article/Use-the-Analysis-ToolPak-to-perform-complex-data-analysis-f77cbd44-fdce-4c4e-872b-898f4c90c007[WebCite
Cache ID 6f3X2Rz37]
11. Microsoft. Data Mining (SSAS) URL: https://msdn.microsoft.com/en-us/library/bb510516.aspx [accessed 2016-02-04]
[WebCite Cache ID 6f3XbsRvU]
12. Waikato. WEKA 3: Data Mining Software in Java URL: http://www.cs.waikato.ac.nz/ml/weka [accessed 2016-01-02]
[WebCite Cache ID 6f3XL1Psy]
13. County Health Rankings. County Health Rankings & Roadmaps: Building a Culture of Health, County by County URL:
http://www.countyhealthrankings.org/rankings/data/fl [accessed 2015-05-20] [WebCite Cache ID 6f3VUACLq]
14. UCI. Heart Disease Data Set URL: https://archive.ics.uci.edu/ml/datasets/Heart+Disease [accessed 2016-01-02] [WebCite
Cache ID 6f3WAvzRv]
15. Hersh W. A stimulus to define informatics and health information technology. BMC Med Inform Decis Mak 2009 May
15;9:24 [FREE Full text] [doi: 10.1186/1472-6947-9-24] [Medline: 19445665]
16. Chambers JM. Greater or lesser statistics: a choice for future research. Stat Comput 1993 Dec;3(4):182-184. [doi:
10.1007/BF00141776]
17. Witten I, Frank E, Hall M. Introduction to data mining. In: Data Mining: Practical Machine Learning Tools and Techniques.
3rd edition. Burlington, MA: Morgan Kaufmann-Elsevier; 2011.
18. Donoho D. 50 Years of Data Science. Csail.mit. 2005 Sep 18. URL: http://courses.csail.mit.edu/18.337/2015/docs/
50YearsDataScience.pdf[WebCite Cache ID queryphp]
19. Jiawei H, Micheline K, Jian P. Data preprocessing. In: Data Mining: Concepts and Techniques. 3rd edition. Waltham, MA:
Morgan Kaufmann-Elsevier; 2012.

Abbreviations
ANOVA: Analysis of variance
IBM: International Business Machines Corporation
IBMWA: IBM Watson Analytics
LR: Logistic regression
NLP: natural language processing
SPSS: Statistical Package for the Social Sciences
WEKA: Waikato Environment for Knowledge Analysis

http://publichealth.jmir.org/2016/2/e157/ JMIR Public Health Surveill 2016 | vol. 2 | iss. 2 | e157 | p. 11

(page number not for citation purposes)
XSL• FO
RenderX
 JMIR PUBLIC HEALTH AND SURVEILLANCE Hoyt et al

Edited by G Eysenbach; submitted 29.03.16; peer-reviewed by S Bhattacharya, CH Li, C McGregor, A Ramachandran; comments to
author 25.07.16; revised version received 31.08.16; accepted 17.09.16; published 11.10.16
Please cite as:
Hoyt RE, Snider D, Thompson C, Mantravadi S
IBM Watson Analytics: Automating Visualization, Descriptive, and Predictive Statistics
JMIR Public Health Surveill 2016;2(2):e157
URL: http://publichealth.jmir.org/2016/2/e157/
doi: 10.2196/publichealth.5810
PMID: 27729304

©Robert Eugene Hoyt, Dallas Snider, Carla Thompson, Sarita Mantravadi. Originally published in JMIR Public Health and
Surveillance (http://publichealth.jmir.org), 11.10.2016. This is an open-access article distributed under the terms of the Creative
Commons Attribution License (http://creativecommons.org/licenses/by/2.0/), which permits unrestricted use, distribution, and
reproduction in any medium, provided the original work, first published in JMIR Public Health and Surveillance, is properly
cited. The complete bibliographic information, a link to the original publication on http://publichealth.jmir.org, as well as this
copyright and license information must be included.

http://publichealth.jmir.org/2016/2/e157/ JMIR Public Health Surveill 2016 | vol. 2 | iss. 2 | e157 | p. 12

(page number not for citation purposes)
XSL• FO
RenderX