Exploratory Data Analysis Of Training Load Metrics In Division I Football Players

Advancement of athlete monitoring technologies has aided training load optimization. However, further development of analysis techniques is needed to effectively interpret and communicate the most meaningful outcomes. PURPOSE: To examine positional training demands of collegiate football players via exploratory analysis of training load metrics. METHODS: Data collected from eight NCAA Division I football players were analyzed by position (2 backs, 6 linemen). Catapult global positioning trackers were used to measure training load during 10 sessions of Fall training camp. Dual Energy X-ray Absorptiometry (DXA) scans were conducted within 3 weeks of training camp. Principal component analyses (PCA) were performed to model position-specific (model 1: backs; model 2: linemen) demands and contained measures of DXA, contact and player loads, change of direction (CoD), acceleration (acc), deceleration (dec), velocity, and session rating of perceived exertion (sRPE). Principal components (PCs) were retained if Eigenvalues were greater than 1.0. RESULTS: Mean body weight was 138.5 ± 14.8 kg for linemen (98.9 ± 6.9 kg lean mass; 25.4 ± 6.7% body fat) and 93.5 ± 2.7 kg for backs (78.0 ± 4.2 kg lean mass; 14.0 ± 0.1 % body fat). Due to collinearity violations, velocity and dec metrics were removed from the PCA. Model 1 was based on 2 PCs, accounting for 57.3% of data variability. The first PC largely represented negative factor loadings of body fat %, and positive factor loadings of lean mass and maximum acc. The second PC demonstrated larger contributions from total player load, CoD and acc. Model 2 identified 3 PCs, accounting for 65.7% of data variability. Each of the 3 PCs reflected external training loads (measures of acc, CoD, and player and contact loads), internal training loads (sRPE) plus lean mass, and body fat %, respectively. CONCLUSION: Statistical modeling may be useful in standardizing and identifying meaningful training load metrics. Athlete monitoring systems supply an array of metrics, however, maximum acc, player load, CoD, and body composition provide beneficial insights into training loads. Using PCs can facilitate a more concentrated and efficient approach to athlete monitoring but may need to be tailored by position.