4 95 TRANSFERRING TARGETED MAXIMUM LIKELIHOOD ESTIMATION INTO SPORT SCIENCE Table 3. Observed dataset causal model. Measure GLM TMLE TMLEH Correct causal model ATE 0.0105 0.0149 0.0142 Confidence Interval 95% -0.0007-0.0216 0.0007-0.0290 -0.0021-0.0303 Miss-specified causal model ATE 0.0193 0.0245 0.0247 Confidence Interval 95% -0.0007-0.0216 0.0115-0.0374 0.0210-0.0381 Difference correct causal model and miss-specified causal model Difference correct causal model and miss-specified 0.0089 0.0096 0.0121 Difference correct causal model and miss-specified % 84.7 65.0 66.3 GLM = Generalized Linear Model; TMLE = Targeted Maximum Likelihood Estimation; TMLEH = Targeted Maximum Likelihood Estimation using Handpicked algorithms; ATE = Average Treatment Effect (i.e., Effect Size) of a substitute in a previous period on the total distance of a soccer team. Figure 5. The Average Treatment Effect of the simulation of the causal model and the missspecified causal model. True ATE = True Average Treatment Effect (i.e. Effect Size) of a substitute in a previous period on the total distance of a soccer team. , CI 95% = Confidence Interval 95%; GLM miss = Generalized Linear Model with miss-specified causal model; GLM = Generalized Linear Model, TMLE miss = Targeted Maximum Likelihood Estimation with miss-specified causal model; TMLE = Targeted Maximum Likelihood Estimation; TMLE hand miss = Targeted Maximum Likelihood Estimation using Handpicked algorithms with miss-specified causal model; TMLE hand = Targeted Maximum Likelihood Estimation using Handpicked algorithms
RkJQdWJsaXNoZXIy MjY0ODMw