Construction projects frequently encounter challenges such as time delays, which can
lead to uncertainty in project completion. The commonly used S-curve control
method only compares planned and actual progress without predicting when the
project will be completed. This is what the forecasting method tries to overcome with
performance indicators and prediction indicators in their concepts. This study
compared these two indicators in measuring performance and predicting completion
time. The forecasting methods employed were Earned Value Management (EVM),
Earned Schedule Management (ESM), and Kalman Filter Earned Value Method
(KEVM). The comparison was conducted using three prediction measurement
criteria: accuracy, precision, and stability, across seven percent completion ranges of
project completion. The data used was from the Banjarharjo Type C Terminal
Improvement Project, which was completed earlier than planned, with a planned
duration of 22 weeks and a final duration of 22 weeks less than 3 days. The results
of the study indicated that the comparison of schedule performance indicators
between EVM and ESM showed that at the end of the 22nd week, both methods
yielded the same results, the value is zero. However, since Schedule Variance
(SV(T)) was reported in time units, SV(T) more accurately reflected the actual
project condition in terms of schedule variance (in weeks). The comparison of
predicted completion times at the end of the 22nd week discovered EVM at 22,000
weeks, ESM at 22,000 weeks, and KEVM at 22,420 weeks. The comparison of
predicted completion times across seven percentage ranges of project completion
revealed that the KEVM method was more accurate, with the lowest Mean Absolute
Error (MAE) and Mean Absolute Percentage Error (MAPE) in six of the completion
percentage ranges. At the same time, EVM showed the lowest MAE and MAPE in
the 75% - 100% range. In terms of precision, KEVM was the most precise, with the
highest average Tx across all seven completion percentage ranges. KEVM also
demonstrated greater stability, with the lowest coefficient of variation (CV) in six
completion percentage ranges, while ESM obtained the lowest CV in the mid-range
of 35% - 70% completion.