Extensive measurements of radon and radon daughters concentration gauge in a radon chamber were carried out. Count rate "spectra" against time at the output of radiation detectors were measured and registered. The count rate spectra were then processed employing Principal Component Regression (PCR). A root mean square error of the count rate was estimated. It was found that PCR processing removes a great part of count rate random fluctuations originating from the radiation statistics that results in a decrease of count rate random error. The root mean square error of count rate in a radon daughter monitor is about 3 times lower, which is equivalent to the error of the gauge with a 9 times higher air flow rate if no PCR processing is used. In case of the radon concentration gauge the increase of sensitivity is even higher and amounts to 5 times.