Evaluating applicability of Google Trend Score in predicting yard waste collection

This study investigates whether Google trends score (GTS) can serve as a useful predictor of yard waste collection. To examine this, yard waste collection patterns in Winnipeg, Canada, from 2010 to 2024 were analyzed, and two multiple linear regression (MLR) models were developed. The study revealed a positive trend (Slope= 1.80, R 2 = 0.37) in yard waste per thousand capita (YPC) and a comparable rise in yard waste pickup per thousand capita (PPC). The variations were influenced by precipitation, vegetation and storm incidents. The social and climatic parameters were used to predict yard waste collection. Pearson correlation (|r| < 0.95, p < 0.05) revealed 5 potential predictors, including average relative humidity (ARH), median income (MI), education attainment (EA), GTS, and average wind speed (AWS). The MLR model containing GTS performed better, achieving an R 2 of 0.72 with all predictors remaining statistically significant (p < 0.05). GTS increased R 2 by 0.17, reduced root mean square error (RMSE) by 1.67, and increased F-statistic by 7.91. Diagnostic tests showed no multicollinearity (Variance inflation factor, VIF=1.25) and no heteroscedasticity (Non-constant variance, NCV p > 0.05). The residuals followed normal distribution (Anderson-Darling p > 0.05). The results show that adding GTS produces an observable improvement in the prediction accuracy of the YPC. It demonstrated that online search interest can serve as a meaningful proxy for public awareness.