Skip to main content
eScholarship
Open Access Publications from the University of California

Recalibration of the complaint prediction model

  • Author(s): Federspiel, C.;
  • Martin, R.;
  • Yan, H.
  • et al.
Abstract

This paper describes the evaluation and recalibration of the complaint prediction model developed by Federspiel (2000). We collected temperature time-series data and complaint data from six buildings ranging in size from 60,000 ft2 to 800,000 ft2 from three different geographical locations. Using these data, we found a low correlation between the observed number of complaint events and the Predicted Average Complaint Events (PACE) for the monitoring intervals and systematic underprediction of hot complaints. We recalibrated the model, increasing the correlation coefficient between observed number of complaint events and PACE to r = 0.49. This degree of correlation, though still not high, is statistically significant (p = 0.044). The recalibrated model predicts that the temperature corresponding to the minimum number of complaints is lower than that of the original model. The recalibrated model also predicts that the minimum number of complaints is greater than that of the original model. Finally, the recalibrated model is not symmetrical. The recalibrated model predicts that hot complaints will increase faster as the average temperature rises than will cold complaints as the average temperature decreases. We used complaint temperatures and an observed setup in building-wide mean temperature to validate the recalibration. From observed complaint temperatures, we constructed six hypothesis tests on predicted values of the mean and standard deviation of complaint temperatures. The differences between the predicted and computed complaint temperature statistics were not statistically significant in all six cases. We compared the observed effect of raising the mean temperature 3°F with the predicted effect. The observed hot complaint rate during the high-temperature period was 2.4 times higher than during the low-temperature period. The predicted ratio was 5.3 times. The difference was explained by underreporting observed by the chief engineer. We expected a dependence of the mean complaint levels on mean outdoor temperature because correlations between mean outdoor temperature, clothing insulation, and indoor air velocity have been established. However, we did not find such an influence. The complaint model predicts that the mean temperature for minimizing complaint rate on arrival is lower than for minimizing complaint rate during the occupied period of the day. This can be explained by a higher metabolic rate on arrival.

Main Content
For improved accessibility of PDF content, download the file to your device.
Current View