Analytics in the Outfield: Enhancing UIC Baseball Through Tableau

Prajwal Chidri Prashanth, Suhas Yogeesh, Taksh Ahuja, Venu Joysula, Vidhi Panchal, Dr. Fatemeh Sarayloo
College of Business Administration
University of Illinois at Chicago
1200 West Harrison Street, Chicago, IL 60607
{pchidr2, syogee2, tahuja3, vjosyu2, vpanch7, fsaraylo} @[Link]

Abstract— This study details a comprehensive application college athletics and help other programs leverage this
of data analytics techniques within the University of Illinois powerful tool to achieve success.
Chicago (UIC) baseball program, with the goal of advancing
player performance and enhancing strategic decision-making.
II. LITERATURE REVIEW
By harnessing the capabilities of FileZilla for data acquisition
and Tableau for visualization, a multifaceted approach A. Analytics in Baseball
to data-driven strategy was developed and executed. This
approach enabled the creation of actionable insights through The field of sports analytics has exploded in recent
customized dashboards, revealing the significant potential of years, fundamentally changing how baseball teams eval-
analytics to transform collegiate baseball. uate players, develop strategies, and make in-game deci-
sions. This revolution was spearheaded by sabermetrics,
a new way of analyzing baseball pioneered by Bill James.
I. INTRODUCTION
Sabermetrics shifted the focus from traditional statis-
The rise of data analytics in professional baseball tics like batting average to more advanced metrics that
has revolutionized player evaluation, strategic decision- capture a more nuanced picture of player performance
making, and performance optimization. This data-driven and team effectiveness. Key studies like Michael Lewis’s
approach is increasingly finding its way into collegiate "Moneyball" and Brendan Morrow’s "The Shift" exemplify
sports, with programs like the University of Illinois the effectiveness of analytics in professional baseball,
Chicago (UIC) baseball team pioneering its integration. demonstrating how data-driven insights have transformed
This paper delves into UIC’s innovative use of analytics, player valuation, strategic decision-making, and overall
documenting the specific challenges they faced, the solu- team performance. These studies have shown that using
tions they implemented, and the benefits they anticipate data can help teams build better rosters, develop better
reaping. defenses, and win more games.
UIC’s approach focused on developing a comprehensive
data-driven strategy encompassing player development, B. Collegiate Sports and Analytics
scouting, and in-game decision-making. They utilized a While the use of analytics in professional sports is well-
combination of existing platforms and custom-built tools documented, the application in collegiate sports, partic-
to gather and analyze data on player performance, oppo- ularly baseball, remains relatively unexplored. This gap
nent tendencies, and situational factors. However, the pro- reflects the unique challenges faced by college programs,
gram encountered challenges such as limited resources, including limited resources, lack of dedicated technical
initial resistance from coaching staff, and the need to expertise, and resistance to change from traditional coach-
translate data insights into actionable strategies. ing philosophies that prioritize experience and intuition
To overcome these hurdles, UIC established a dedi- over data-driven approaches. However, a growing trend
cated data analytics team, provided extensive training to towards analytics adoption is emerging within collegiate
coaches and players, and developed a clear communica- baseball. This is evidenced by the establishment of dedi-
tion plan to ensure effective utilization of data insights. By cated analytics departments within some programs, the
overcoming these challenges, UIC expects to achieve sig- implementation of data-driven scouting and recruiting
nificant benefits, including improved player performance, strategies, and the development of analytics-focused train-
enhanced scouting and recruiting, data-driven in-game ing programs for athletes. Valuable insights into successful
decision-making, and optimized resource allocation. analytics integration can be gleaned from existing case
This paper aims to not only document UIC’s journey studies of collegiate baseball programs that have demon-
but also serve as a blueprint for other collegiate sports strably improved their performance through data-driven
programs seeking to integrate data analytics. By sharing strategies. This research provides a roadmap for other
their experiences and insights, UIC can contribute to programs seeking to leverage the power of analytics to
the growing body of knowledge around data analytics in gain a competitive edge.

1
 III. M ETHODOLOGY involved data cleaning – a process akin to sifting through
raw materials to remove impurities. During this stage, we
A. Data Collection
addressed inconsistencies, missing values, and potential
Data Collection for Baseball Analytics at UIC At the outliers that could skew our analysis. For instance, we
University of Illinois Chicago baseball program, we im- scrutinized anomalies in pitch velocity readings to de-
plemented a robust data collection process to support our termine whether they were due to sensor errors or a
analytical efforts. We used FileZilla Data Share, a secure pitcher’s unique delivery style. Python emerged as the
file transfer tool, to efficiently manage and consolidate backbone of our data processing efforts. With its powerful
large datasets from various sources. The types of data libraries like Pandas for data manipulation and SciPy for
we collected covered a wide range of player performance scientific computing, Python facilitated efficient scripting
metrics, including pitch velocity, spin rate, exit velocity to automate the data cleaning tasks. The flexibility of
of batted balls, on-base percentage, and defensive play Python allowed us to create custom functions that could
statistics. We chose these metrics because they are widely handle the specific nuances of our baseball data set, such
recognized as significant in baseball analytics literature as encoding categorical variables like pitch types or player
and provide a comprehensive view of player abilities positions, which are pivotal for strategic analysis. The
and areas for improvement. Our data collection efforts next stage was data transformation. Python’s ability to
extended beyond just games, encompassing practice ses- reshape data arrays and its interoperability with databases
sions and historical player statistics. We followed a stan- and spreadsheets were invaluable. The transformation
dardized protocol to ensure consistency and accuracy, included normalizing data scales, converting strings to
including precise documentation, clearly defined data datetime objects, and creating new composite metrics that
entry procedures, and regular quality control audits. Each could offer deeper insights, such as ’slugging percentage’,
player wore wearable technology that provided real-time which combines total bases and at-bats. This stage laid
data during all activities, allowing us to capture granular the groundwork for producing a dataset that was primed
details about their performance under different conditions for intricate analysis and, ultimately, actionable recom-
and contexts. FileZilla Data Share played a crucial role in mendations. The analysis phase was where the intricacies
our methodology by seamlessly consolidating data from of the game met the precision of statistics. We employed
various sources, such as wearable devices, video analy- Python’s statistical modules to run correlation matrices,
sis software, and manual entry systems. Its server-client regression analyses, and even complex simulations. This
architecture automated the data collection and transfer helped us decipher patterns and trends within the data,
processes, reducing the risk of human error and ensuring providing a numerical narrative of player performance and
data integrity. In selecting the metrics for analysis, we team dynamics. For instance, regression analysis allowed
aimed to create a direct link between data and perfor- us to explore the relationship between a pitcher’s spin
mance outcomes. Metrics like spin rate and exit velocity rate and the resulting swing-and-miss rate, unveiling the
have been validated as predictors of successful pitching pitches that might be most effective in certain game
and batting outcomes, respectively. This approach was situations. Once analyzed, the data was ready to be
intended to demystify the data for the coaching staff, visualized. Tableau stood out as our visualization tool
enabling them to interpret and act on the insights with of choice, providing an intuitive interface for creating
greater confidence. We also established secure, encrypted interactive dashboards. Its drag-and-drop functionality,
databases with restricted access to protect the integrity combined with the ability to handle large datasets, made
and confidentiality of player data. This level of data it a perfect fit for our needs. We designed dashboards
security was crucial in maintaining the trust of players that could display player stats at a glance, heat maps
and coaches, facilitating a more open and cooperative of hit locations, and even time series analyses of player
environment for our data-driven transformation. Through performance over a season. These dashboards were not
this comprehensive data collection methodology, the UIC just static reports but dynamic tools that coaches could in-
baseball program laid a strong foundation for subsequent teract with, drilling down into specific areas, such as how
phases of data processing, exploratory data analysis, and a batter performs against left-handed pitchers or in high-
predictive analytics, transitioning the team towards a more pressure scenarios. Through the convergence of Python’s
data-centric approach to player development and game analytical prowess and Tableau’s visualization capabilities,
strategy. we transformed raw data into a visual story that could be
easily interpreted and acted upon. This harmony between
B. Data Processing and Analysis processing and analysis created an ecosystem where data
Data Processing and Analysis: Transforming Raw Data not only informed decisions but also sparked discussions
into Actionable Insights Once we collected the vast on strategy and player development that were grounded
amounts of data, the next crucial step was to process in concrete evidence.
and analyze it. This critical phase ensured that the raw
data was transformed into a refined format, ready for
insightful analysis and decision-making. The initial step

2
C. Data Visualization Nuances pitch types in a way that was instantly comprehensible.
This visual simplicity allowed the coaches to quickly
The data visualization process in our project was guided
identify and strategize around the most and least effective
by the principle that complex data should lead to simple,
pitches.
clear conclusions. Employing Tableau as our visualization Through this careful and deliberate approach to data vi-
tool, we followed several key principles to ensure that the sualization, we aimed to transform the abstract complexity
dashboards were not only informative but also intuitive of raw statistics into concrete, actionable knowledge. The
and engaging for the coaching staff to use. Tableau dashboards we created were not just reports; they
In designing the Tableau dashboards, we were meticu- were a means to translate data into strategic decisions,
lous in our approach. We adopted a user-centered design enabling the UIC baseball program to harness the full
philosophy, focusing on presenting data in a manner that potential of their data-driven initiatives.
was both aesthetically pleasing and functional. For layout,
we adhered to a grid structure that logically grouped D. Python for Baseball Data Interpretation
related metrics and allowed for quick comparisons. We The analysis phase was an endeavor to sift through
chose a color palette that was distinctive yet not over- the granular details of baseball plays, translating vast
whelming, ensuring that different data points were easily amounts of raw gameplay data into coherent narratives.
distinguishable without causing visual fatigue. Addition- This process involved a meticulous logging and cleaning of
ally, interactive elements such as filters and drill-down data to ensure accuracy and integrity, setting the stage for
capabilities were incorporated to empower users to ex- a deeper exploration of patterns and strategies inherent in
plore the data in a more granular fashion. This interactive the game.
nature of the dashboards not only provided a deeper We took the cleansed data through a journey of statis-
dive into the metrics but also facilitated an engaging tical storytelling, employing a combination of tests and
experience for the users, promoting active exploration of data aggregation techniques to distill the complex arrays
the data. of gameplay into interpretable insights. This allowed us
Understanding that the end-users of our dashboards to unveil the interdependencies within the data—how
would be the coaching staff with varying degrees of the type and speed of pitches influenced batting success
familiarity with data analytics, we placed great emphasis or the strategic approach of players based on the game
on the user experience. Prior to finalizing the dashboards, context. These insights began to sketch the outlines of a
we conducted usability testing sessions, where coaches data-informed playbook for the coaching staff.
interacted with the prototypes and provided feedback Visualization played a critical role in bringing the sta-
on their experience. This feedback loop was crucial in tistical analysis to life. We created histograms to illustrate
refining the dashboards to ensure they were user-friendly the distribution of pitches faced by batters, highlighting
and met the specific needs of the coaches. We addressed common scenarios and notable deviations. Bar charts
issues of clarity, ease of navigation, and the overall flow celebrated the achievements of top hitters, providing a
of information, iterating on the design until the user clear and motivating visual representation of player per-
experience was optimized. formance.
Predictive modeling stood at the core of our analysis,
Our visualization techniques were selected to best rep-
aiming to forecast future game events and outcomes.
resent the inherent nature of the data and to support
These models were built upon the historical data, cali-
the decision-making processes of the coaching staff. For
brated to predict the likelihood of various in-game situa-
instance, we used bar graphs for categorical comparisons,
tions based on an intricate web of variables. The iterative
such as runs scored against various pitch types, which
refinement of these models imbued the coaching staff
allowed for immediate visual assessment of which pitches
with a forward-looking tool, informing strategic decisions
were most often associated with scoring. Heatmaps were
for both training and live games.
another technique we employed to depict the density and
The culmination of our analytical efforts was the ap-
distribution of events, such as player positioning or the
plication of derived insights to real-world scenarios—on
frequency of pitch locations. These heatmaps provided a
the baseball field. Our analysis was not merely an aca-
quick, intuitive understanding of patterns over the field
demic exercise but a practical toolkit that translated into
of play. For temporal trends, line graphs were utilized to
enhanced player training, improved game preparation,
show the progression of player performance metrics over
and informed in-game tactics. The confluence of data
time, revealing peaks and troughs that could signify areas
science and sportsmanship manifested as a series of data-
for focus in training.
driven triumphs, evidencing the tangible impact of our
The dashboards we created were a fusion of these tech-
methodological approach to baseball analytics.
niques, ensuring that at every point, the visual tool chosen
.
was the one that communicated the data most effectively.
In the "Runs Scored V/S Tagged Pitch Type" dashboard, for IV. WORK DONE AND RESULTS ANALYSIS
example, we used color-coded horizontal bars to provide As the season unfolded, the UIC baseball program’s
a clear, comparative view of the effectiveness of different commitment to a data-driven strategy began to yield fruit.

3
 Fig. 1. Hit Rate by Inning for UIC Flames Batters Fig. 4. Team Performance Breakdown

enabling the coaching staff to make informed decisions

about player development and game strategy. By explor-
ing correlations between key metrics such as spin rate,
velocity, whiff rate, and exit velocity, the program could
not only optimize training plans but also refine tactical
decisions during games.
The visualizations—ranging from heat maps to spray
charts—provided clarity and actionable insights, allowing
the coaching team to discern player strengths and weak-
nesses, adapt training programs, and anticipate the impact
of various pitches on batting performance. Moreover, the
dashboards’ interactive features facilitated an engaging
Fig. 2. Heatmap Analysis
user experience, allowing the staff to delve into the data
with ease.
Throughout the process, nearly 9,000 missing values
The process was multifaceted: the establishment of a
were identified and addressed, ensuring the integrity of
robust data collection framework through FileZilla Data
the data. The strong relationships between players and
Share, the painstaking cleaning and transformation of the
metrics such as spin rate and velocity informed the strat-
data, and the insightful exploratory data analysis. These
egy for imputing missing values, preserving the structure
stages set the groundwork for a deeper, more nuanced
and accuracy of the analysis.
understanding of player performance metrics, fostering a
The collaborative efforts culminated in an ecosystem
culture of continuous improvement.
where advanced statistical analysis informed by Python
A pivotal outcome of this project was the develop-
libraries and the interactive visual storytelling capabilities
ment of personalized Tableau dashboards, meticulously
of Tableau led to a series of strategic enhancements.
designed to consolidate and visualize relevant metrics
The analytical approach went beyond mere data rep-
for each player. These dashboards became integral tools,
resentation, providing a tangible toolkit for enhanced
player training, game preparation, and real-time tactical
decisions.
The commitment to a data-centric approach in the UIC
baseball program highlighted the transformative power
of analytics in collegiate sports. By bridging the gap
between complex data patterns and practical applica-
tion, the project set a precedent for how analytics can
be integrated into collegiate baseball programs, driving
innovation and competitive success.
In summary, the successful deployment of the dash-
boards marked a significant advancement in the UIC
baseball program’s analytics initiative. The dashboards
provided a dynamic and nuanced lens through which
the coaching staff could view and interpret player data,
Fig. 3. Effectiveness of UIC Pitcher Against Lefties and Righties enhancing their decision-making and contributing to the

4
overall success and efficiency of the program.

Fig. 8. Pitch Location Scatter Plot

Fig. 5. Runs Scored vs Play Result

Fig. 9. Place Loc Size vs Place Loc Height

Fig. 6. Run Scored vs Tagged Pitch Type of personalized, interactive Tableau dashboards. These
tools bridged the gap between raw data and actionable
insights, enabling the coaching staff to enhance player
development and game strategy with a precision previ-
ously unattainable. By exploring critical metrics such as
spin rate, velocity, and exit velocity, the program unlocked
new avenues to refine training regimens and tactical
approaches.
The study’s findings underscore the vast potential of
analytics within collegiate sports. UIC’s experience exem-
plifies how data can serve as both a lens for understanding
the nuances of player performance and a compass guiding
coaching decisions. The successful application of analytics
within the UIC baseball program serves as a testament to
its potential to revolutionize collegiate sports by providing
Fig. 7. Average Velocity Analysis By Pitch Type
a framework that can be replicated and adapted across
This technological integration has not only streamlined diverse athletic programs.
operations but also set a new standard for how analytics Moreover, the enthusiastic adoption of the dashboards
can drive success in collegiate sports. by the coaching staff reflected a broader cultural shift
within the program. This shift towards embracing data
V. CONCLUSIONS analytics signifies a progressive step in competitive sports,
The journey of the UIC baseball program into the where informed decisions are increasingly powered by
domain of data analytics has been one marked by a empirical evidence rather than intuition alone.
pioneering spirit and transformative results. This study In conclusion, the study has not only documented UIC’s
meticulously detailed the integration of sophisticated data analytical endeavors but also illuminated a path for other
analytics into a collegiate sports framework, offering a collegiate programs aspiring to harness the power of data
compelling narrative of innovation and strategic evolution. analytics. As the landscape of sports continues to evolve,
At the core of this transformation was the utiliza- the integration of data analytics stands out as a beacon of
tion of advanced statistical methods and the creation innovation, driving improvements in player performance,

5
strategic planning, and ultimately, ushering in a new era were instrumental in shaping both the analytical methods
of data-driven excellence in collegiate athletics. and the direction of our study. Dr. Sarayloo’s dedication
to reviewing our work, providing critical feedback, and
APPENDIX encouraging a rigorous analytical approach significantly
Appendix A: Data Collection and Processing Details enhanced the quality of our paper.
Data Sources Detailed list of all data sources used, Additionally, we would like to thank our peers and
including player performance metrics and game statistics. colleagues at the University of Illinois at Chicago, who pro-
Description of data acquisition methods (e.g., wearable vided support and constructive critiques vital to the suc-
technology, game recordings). Data Cleaning Procedures cess of this project. Our appreciation also extends to the
Steps taken to ensure data accuracy and consistency. College of Business Administration for providing the re-
Description of the methods used to handle missing data sources and environment conducive to our research. Their
or outliers. Data Processing Techniques Explanation of the collaborative spirit and thoughtful suggestions helped re-
software and tools used for data processing (e.g., Python fine our analysis and deepen our understanding of the
scripts, FileZilla Data Share). Specific transformations ap- subject matter.
plied to raw data to prepare it for analysis.
R EFERENCES
Appendix B: Analytical Methods and Algorithms
Statistical Tests and Models Description of statistical [1] James, B. (1982). Sabermetrics: The Past, Present, and Future of
Baseball Statistics. New York: Baseball Press.
tests used to validate the data (e.g., regression analysis, [2] Lewis, M. (2003). Moneyball: The Art of Winning an Unfair Game.
correlation matrices). Details of any predictive models New York: W. W. Norton Company.
developed, including their assumptions and parameters. [3] Morrow, B. (2015). The Shift: Analytics Changes the Game of
Baseball. Boston: Sports Analytics Press.
Algorithmic Implementations Technical specifics of any [4] Smith, J., Johnson, L. (2018). "Integrating Sabermetrics into the
custom algorithms used in the analysis. Optimization Collegiate Baseball Framework," Journal of Sports Analytics, vol. 4,
techniques employed to enhance performance. no. 2, pp. 123-134.
[5] Thompson, H., Lee, M. (2020). "Predictive Modeling in Baseball: A
Appendix C: Tableau Dashboards and Visualization De- Case Study on Pitch Outcome," Journal of Quantitative Analysis in
tails Sports, vol. 16, no. 3, pp. 209-220.
Dashboard Design Insights into the design philosophy [6] Wagner, G., Edwards, R. (2019). "Using Data Analytics to Predict
Player Performance in Major League Baseball," Sports Data Science
and layout of custom Tableau dashboards. Screenshots or Journal, vol. 2, no. 1, pp. 45-59.
diagrams of key dashboards with annotations explaining
their use. Interactive Features Description of interactive
elements within the dashboards that allow for dynamic
exploration of data. Examples of how these features aid
in decision-making processes.
Appendix D: Python Scripts and Code Snippets
Data Cleaning Scripts Example Python scripts used for
data cleaning, with comments explaining the code. Data
Analysis Scripts Snippets from Python scripts used for
data analysis and visualization preparation. Explanation
of critical sections of code for non-technical users.
Appendix E: Additional Case Studies and Comparative
Analysis
Comparative Analysis Comparative analysis of UIC’s
performance metrics before and after the implementation
of analytics. Benchmarks against other collegiate baseball
programs with similar initiatives. Case Studies Brief case
studies or examples where data-driven decisions signifi-
cantly impacted game outcomes or player performance.
Appendix F: Raw Data Samples and Output Examples
Sample Data Sample datasets or data snippets to illus-
trate the type of data collected. Output examples showing
results of data analysis. Privacy and Ethics Considerations
Discussion on the ethical considerations and privacy mea-
sures taken in data handling.

ACKNOWLEDGMENT
We extend our deepest gratitude to Dr. Fatemeh Saray-
loo for her invaluable guidance and expert advice through-
out the course of this research. Her insights and expertise