-0.85
Visit Frequency Impact
Visit frequency is the strongest predictor of churn with the highest negative correlation to member retention
100%
Prediction Accuracy
Model predicted every case correctly with no false alarms on this exceptionally clear dataset
30x
Behavior vs Age Impact
Training habits are 30 times more powerful than age in predicting which members will quit
>0.50
High-Risk Threshold
Members with risk scores above 0.50 need immediate encouragement and support to stay engaged
Member Behavior Analysis Insights
Attendance is the Clearest Warning Sign
Visit frequency stands out as the most meaningful predictor of churn. Despite statistical testing challenges due to complete separation, the real-world signal is clear: members who visit less frequently are far more likely to quit. A steady drop in visits is the earliest and most reliable sign that a member is losing motivation.
Best Predictor Combination
The most effective combination includes visit frequency, workout duration, and calories burned. This mix reflects real member behavior: attendance shows commitment, while workout duration and calories burned reveal engagement during each visit. Age adds little predictive value but helps personalize coaching approaches.
Two-Tier Risk Distribution
Most gym members have extremely low churn risk - they visit regularly, train consistently, and maintain stable habits. A smaller group shows near-maximum risk with rare visits and inconsistent workouts. Without early intervention, this high-risk group is likely to drop off completely.
Statistical Limitations Don't Diminish Real Patterns
The model encountered complete separation - members with very low visits almost always churn while high-frequency visitors never do. This technical issue prevented proper p-value calculation, but the behavioral patterns remain exceptionally strong and meaningful for practical trainer decision-making.
Data Visualization Insights
Correlation and Multicollinearity Analysis
Analysis of how different member behaviors relate to churn risk. Visit frequency shows the strongest negative correlation (-0.85) with retention, while workout duration, calories burned, and age have minimal overlap, making them complementary predictors.
