Simulation dataset on power and size of Wald, likelihood-ratio, and score test statistics in parametric competing risks models under hybrid censoring

This dataset contains Monte Carlo simulation results for parametric competing risks models (CRM) with K=2 or K=3 independent competing causes of failure under hybrid (Type I/II combined) censoring. The dataset was generated as part of the doctoral research at the National University of Uzbekistan. The archive includes 13 CSV files covering: - Bias and RMSE of the maximum likelihood estimator (MLE) for exponential, Weibull (shape alpha=2), and Gompertz families; - Empirical Type I error and power of Wald (Wn), likelihood-ratio (LRn), and score (Sn) test statistics at nominal level alpha=0.05; - Non-centrality parameter lambda* and theoretical power under Pitman local alternatives; - Coverage and mean length of asymptotic and parametric bootstrap 95% confidence intervals; - Robustness to risk dependence via Clayton copula (theta_cop in {0, 0.3, 0.5, 1.0}); - Minimum required sample size n* for a given precision target; - Application to the publicly available lung cancer dataset (survival::lung, n=228). Experimental grid: n in {50, 100, 200, 500}, beta in {0.25, 0.50, 0.75, 1.00}, R=5000 replications per setting. Python 3.11, NumPy 1.24, SciPy 1.11; R 4.3.2 (survival, cmprsk). Base random seed 42 with independent SeedSequence streams per parallel worker.