To analyze the main causes of bad air quality in recent years and the relationships embedded between air pollutants, this paper uses an improved Apriori algorithm to analyze the data mining of the correlations between overall air quality and air pollutants and between air pollutants in Guangzhou city's air quality data from 2014 to 2020.The data are divided according to AQI levels and air pollutant exceedance thresholds, and a Boolean type transaction dataset is constructed. Based on this, support, confidence and boosting thresholds are set to mine the transactional dataset for frequent item sets and obtain strong association rules. The results show that (1) The excess concentrations of PM10, CO, and NO2 play a role in the excess concentrations of PM2.5. (2) In recent years, the main pollutant indicators of air pollution are O3, NO2, PM2.5, PM10, among which O3 has the greatest influence. The strong correlation rules obtained through data mining have some significance to the detection and prevention of air pollution.