Abstract: Based on the parallel observational data from the DDL-2 and RAD7 radon monitors at Well Anguo-1 in Pingliang, Gansu Province, this study systematically analyzes the influence of different sampling methods on DDL-2 instrument measurements by integrating traditional statistical methods (including Pearson correlation analysis and coefficient of variation calculation) with machine learning algorithms (including random forest regression and Isolation Forest anomaly detection). The research identifies and evaluates the interference of key environmental parameters (temperature, humidity, and atmospheric pressure) on the measurement accuracy of the DDL-2 instrument, and compares the long-term stability between the domestic DDL-2 radon monitor and the imported RAD7 radon monitor. The results indicate that temperature and atmospheric pressure exert relatively significant effects on the measurement precision of the DDL-2 radon values. The adoption of standardized sampling methods and controlled observation conditions can substantially reduce the primary-duplicate sample errors in DDL-2 measurements, with the stability of DDL-2 measurements approaching that of the RAD7 instrument. This study takes Well Anguo-1 in Pingliang, Gansu Province as a case to evaluate the applicability of the DDL-2 radon detector in seismic monitoring, providing a scientific basis for data quality control and standardized operation of domestic radon detectors.