Page 55 -

54 Part I • Decision Making and Analytics: An Overview

Application Case 1.5

Analyzing Athletic Injuries
Any athletic activity is prone to injuries. If the inju- Neural network models were built to pre-
ries are not handled properly, then the team suf- dict each of the healing categories using IBM SPSS
fers. Using analytics to understand injuries can help Modeler. Some of the predictor variables were cur-
in deriving valuable insights that would enable rent status of injury, severity, body part, body site,
the coaches and team doctors to manage the team type of injury, activity, event location, action taken,
composition, understand player profiles, and ulti- and position played. The success of classifying the
mately aid in better decision making concerning healing category was quite good: Accuracy was 79.6
which players might be available to play at any percent. Based on the analysis, many business rec-
given time. ommendations were suggested, including employ-
In an exploratory study, Oklahoma State ing more specialists’ input from injury onset instead
University analyzed American football-related sport of letting the training room staff screen the injured
injuries by using reporting and predictive analytics. players; training players at defensive positions to
The project followed the CRISP-DM methodol- avoid being injured; and holding practice to thor-
ogy to understand the problem of making recom- oughly safety-check mechanisms.
mendations on managing injuries, understanding
the various data elements collected about injuries, Questions for Discussion
cleaning the data, developing visualizations to draw
various inferences, building predictive models to 1. What types of analytics are applied in the injury
analyze the injury healing time period, and drawing analysis?
sequence rules to predict the relationship among the 2. How do visualizations aid in understanding the
injuries and the various body part parts afflicted with data and delivering insights into the data?
injuries. 3. What is a classification problem?
The injury data set consisted of more than 4. What can be derived by performing sequence
560 football injury records, which were categorized analysis?
into injury-specific variables—body part/site/later-
ality, action taken, severity, injury type, injury start What We can Learn from this application
and healing dates—and player/sport-specific varia- case
bles—player ID, position played, activity, onset, and For any analytics project, it is always important
game location. Healing time was calculated for each to understand the business domain and the cur-
record, which was classified into different sets of rent state of the business problem through exten-
time periods: 0–1 month, 1–2 months, 2–4 months, sive analysis of the only resource—historical data.
4–6 months, and 6–24 months. Visualizations often provide a great tool for gaining
Various visualizations were built to draw the initial insights into data, which can be further
inferences from injury data set information depict- refined based on expert opinions to identify the rela-
ing the healing time period associated with players’ tive importance of the data elements related to the
positions, severity of injuries and the healing time problem. Visualizations also aid in generating ideas
period, treatment offered and the associated healing for obscure business problems, which can be pur-
time period, major injuries afflicting body parts, and sued in building predictive models that could help
so forth. organizations in decision making.

prescriptive analytics
The third category of analytics is termed prescriptive analytics. The goal of prescriptive
analytics is to recognize what is going on as well as the likely forecast and make decisions
to achieve the best performance possible. This group of techniques has historically been
studied under the umbrella of operations research or management sciences and has gen-
erally been aimed at optimizing the performance of a system. The goal here is to provide

M01_SHAR9209_10_PIE_C01.indd 54 1/25/14 7:46 AM