Correlation between GPS and wellness metrics over time

Project Background

With GPS technology used in every training session, and wellness questionaires completed daily, sports scientists and high-performance staff must understand how different training loads impact athlete wellbeing. This project aimed to understand the correlation between eight key GPS metrics, the most commonly used for load management, and their effect on athlete wellness ratings both the next day and two days after training.

Athletes report wellness ratings on a 1-7 scale, with 7 indicating the most positive state. For example, a soreness rating of 7 means no soreness, whereas 1 indicates severe soreness. To account for individual differences in perception, these ratings are converted into Z-Scores, ensuring that variations are standardised across the squad. Players also receive guided definitions for each rating to maintain consistency in reporting.

This visualisation offers key takeways into the physiological and psychological impact of training, helping staff understand how athletes are responding and anticipate future trends. It highlights which GPS metrics contribute to fatigue and soreness and how different training loads influence sleep, mood, and stress levels. By leveraging these insights, training loads can be adjusted to optimise recovery, ensuring players stay fresh while maintaining peak performance.

Stakeholders

• High-Performance Manager

• Athletes

• Strength & Conditioning coaches

• Physiotherapists

• Medical Team (Club Doctor, Welfare)

Data Structure

This is a real-world dataset taken from Australian football athletes across an entire season. The ‘gps_cleaned’ dataset contains 1,966 rows × 10 columns, and the ‘wellness_cleaned’ dataset contains 5,502 rows × 11 columns.

Data Cleaning, Transformation & Manipulation

SQL was used to find the correlation between GPS metrics and wellness metrics over one and two days post-training. This process began with data preparation, where GPS and wellness data were joined, incomplete records were excluded, and wellness metrics were aligned with GPS data based on training dates. The Athlete Management System that collects wellness questionnaires allows athletes to modify their responses, with answers autosaving periodically. As a result, multiple entries may be recorded for the same athlete on a given day. However, the Z-Score is only calculated once the form is fully completed and submitted. To maintain data integrity, a filtered dataset was created, ensuring that only records with valid Z-Scores were included.

The data was then reshaped using an unpivoting technique to standardise GPS and wellness metrics into a long format for correlation analysis.

Finally, a correlation calculation was performed using Pearson Correlation Coefficient formula to measure the relationship between GPS and wellness metrics, ranking them in descending order. After data manipulation and transformation, the final dataset contained 144 rows and 3 columns.

Data Visualization

A heatmap was created in Tableau to highlight the correlation between GPS metrics and wellness metrics. A side-by-side comparison was used to illustrate the impact of an additional day of recovery, providing a clearer view of how training loads influence athlete wellbeing over an acute time period.

For this project, the scenario question is used:

"What impact does training have on athlete wellness, and which wellness metrics are most influenced by GPS metrics over an acute time period?"

Overview of findings

Total Distance, Run, High-Speed Running (HSR), and Deceleration have a negative impact on DOMS, Recovery, Readiness, Fatigue, and Sleep the day after training, with these correlations weakening on the second day. While the relationships are weak, they highlight the role of training load in influencing acute recovery in AFL athletes. Conversely, these same GPS metrics show a slight positive association with sleep quality and quantity after two days, likely as a compensatory response to reduced sleep in the nights following a hard training session. Additionally, Deceleration, Sprint, Accelerations, and Sprint Efforts show a mild positive correlation with stress management after two days.

Negative Associated GPS Metrics on Wellness

• The four main GPS metrics of Total Distance, Run, High-Speed Running (HSR), and Deceleration show a negative association with Delayed Onset Muscle Soreness (DOMS), Recovery, Readiness, Fatigue, and Sleep Quantity and Quality the day after training. After two days, these metrics continue to impact the same wellness factors, except for sleep quantity and quality.

• The strongest correlations include Run to Recovery (+1 day) at -0.28, Total Distance to DOMS (+1 day) at -0.28, Total Distance to Recovery (+1 day) at -0.28, and Run to DOMS (+1 day) at -0.27.

• On the second day of recovery, these correlations weaken further, ranging between -0.13 and -0.17.

• Whilst these are weak correlations, given the complex nature of sport and recovery, these metrics are essential to understanding the impact a training session will have on an athlete.

Positive Associated GPS Metrics on Wellness

• The four main GPS metrics of Total Distance, Run, High-Speed Running (HSR), and Deceleration show a slight positive correlation with sleep quantity and quality after two days of recovery, ranging between 0.03 and 0.08.

• Stress also has a mild positive correlation with Deceleration, Sprint, Accelerations, and Sprint Efforts two days after training, with values ranging from 0.05 to 0.07.

• While these correlations are weak, they suggest a potential link between high-intensity efforts and improved stress management after 2 days.

Recommendations

Based on the uncovered insights, the following recommendations have been provided:

• Select the right times to incorporate high volumes of Total Distance, Run, High Speed Running (HSR), and Deceleration, as these metrics impact performance based wellness indicators. Typically, these should be reduced two days before a match to reduce unnecessary fatigue.

• The mild correlation between Deceleration, Sprint, Accelerations, and Sprint Efforts and improved stress management suggests that incorporating repeated high-intensity efforts two days before a match may be beneficial, helping to build confidence in athletes.

• This aligns with common match day preparation, where final training sessions focus on lower volumes with increased intensity to enhance readiness.