Table 6 Accuracies and mean ranks for all models across each year for Clinic 2
From: Forecasting patient flows with pandemic induced concept drift using explainable machine learning
| Â | 2017 | 2018 | 2019 | 2020 | 2021 | 2022 | Mean R | ||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Â | MAPE | R | MAPE | R | MAPE | R | MAPE | R | MAPE | R | MAPE | R | |
kNN | 17.4 | 12.2 | 20.4 | 13.0 | 19.5 | 12.8 | 27.9 | 12.2 | 24.1 | 11.5 | 20.2 | 11.9 | 12.3 |
Benchmark | 18.3 | 11.7 | 16.8 | 10.3 | 17.0 | 11.4 | 28.2 | 9.5 | 29.9 | 12.7 | 20.2 | 10.9 | 11.1 |
SVR (NU) | 15.4 | 11.0 | 19.1 | 11.8 | 18.9 | 12.7 | 28.4 | 10.7 | 22.1 | 9.8 | 16.9 | 9.6 | 11.0 |
Naive | 16.4 | 10.4 | 17.3 | 10.2 | 17.9 | 11.8 | 23.6 | 7.6 | 20.8 | 11.5 | 18.9 | 10.0 | 10.3 |
Prophet | 12.4 | 6.5 | 13.1 | 6.9 | 11.2 | 6.5 | 23.8 | 10.0 | 16.9 | 8.2 | 18.3 | 10.2 | 7.9 |
Kernel Ridge | 13.2 | 8.5 | 13.7 | 7.9 | 11.2 | 7.3 | 17.4 | 7.6 | 14.9 | 7.0 | 13.9 | 6.2 | 7.5 |
Ridge | 13.0 | 7.8 | 13.7 | 7.8 | 11.3 | 7.3 | 17.5 | 7.5 | 14.9 | 6.7 | 13.8 | 6.0 | 7.3 |
ARIMA | 12.3 | 6.3 | 13.8 | 7.4 | 11.0 | 6.2 | 17.4 | 8.3 | 16.4 | 7.4 | 16.1 | 8.0 | 7.2 |
CatBoost | 12.6 | 7.7 | 13.3 | 7.4 | 11.2 | 7.2 | 14.6 | 6.2 | 14.2 | 6.7 | 15.6 | 7.9 | 7.1 |
Gradient Boosting | 12.7 | 7.9 | 13.4 | 6.8 | 11.2 | 7.0 | 15.7 | 7.2 | 14.8 | 6.8 | 14.7 | 6.6 | 7.1 |
Random Forest | 12.4 | 7.4 | 13.4 | 6.9 | 11.3 | 7.4 | 16.5 | 6.6 | 15.2 | 6.9 | 15.1 | 7.2 | 7.1 |
Naive (Enhanced) | 12.2 | 6.5 | 13.3 | 7.1 | 11.6 | 7.1 | 17.3 | 7.6 | 16.4 | 7.5 | 13.4 | 5.3 | 7.0 |
Stacking | 12.0 | 6.2 | 12.6 | 5.7 | 10.3 | 5.2 | 17.9 | 6.8 | 14.5 | 6.3 | 15.5 | 7.6 | 6.2 |
Averaging | 11.6 | 5.2 | 12.6 | 5.5 | 10.2 | 5.1 | 16.7 | 6.1 | 14.2 | 5.8 | 15.1 | 6.8 | 5.6 |
Voting | 11.5 | 4.8 | 12.6 | 5.3 | 10.1 | 5.1 | 16.5 | 6.1 | 14.3 | 5.4 | 14.6 | 5.8 | 5.4 |