I created decision tree, boosted tree, and decision forest models with an 80/20 split of the dataset. There are only 4 attributes and 1 target variable. Evaluations of all three models produced the same exact results - same accuracy, recall, precision, etc at a 50% threshold. What does this mean? Why are all the values the same? Also, why are the false negatives so high?

Hotel_bookings | Training (80%) DUC Hotel_bookings | Test (20%) O Positive class: Yes Probability threshold: 509 Max. phic 0 2704 TP FM FP TH PREDICTEDVS. ACTUAL No PREDICTED PRECISION Phi 1.319 0.24 0.27 Yes 1.195 124 90.60%% 7,562 14,966 72.528 86.43% 0.80 0 27 78 57% 0.53 ACTUAL 8.757 15 090 23 847 AVG PRECISION AVG Phi RECALL 13.65% 66 414 AVG. RECALL 100% 90%% 67.8% 0.2372 Accuracy F-measure 80% 70% 90.6% 13.6% 0.2704 60% Precision Recall Phi coefficient True Positive Rate (TPR) 13.65% 0.8% 5.5% 246.7% FPR % positive instances Lift 30% 20%% 15.7% 0.3243 0.3566 10%% K-S statistic Kendall's Tau Spearman's Rho ROC AUC: 0.6218 10% 20%% 30% 40% 50% 60%% 70%% 80% 90% 100% False Positive Rate (FPR) 0.82%