Table 2 Feature names as they appear in figures and their descriptions
From: Forecasting patient flows with pandemic induced concept drift using explainable machine learning
Feature name | Description |
|---|---|
lag7d | Autoregressive 7-day lag, the value from one week prior |
lag14d | Autoregressive 14-day lag, the value from two weeks prior |
lag1 | Autoregressive 364-day lag, the value from one year prior |
lag2 | Autoregressive 728-day lag, the value from two years prior |
lag3 | Autoregressive 1092-day lag, the value from three years prior |
public_holiday | Public Holiday Indicator (0/1) |
week | Week Number ranging from 1–53 |
ped_count | Auckland CBD foot traffic |
flu_percent | weekly reports on Influenza prevalence in New Zealand |
covid_level | 1–4. Prior to 3/12/2021: COVID-19 Alert Level ranging from 1–4; After 3/12/2021: Traffic Light Mappings (Green = 1; Orange/Red = 2; No mappings to 3 or 4) |
trends | Google Trends normalised frequency of term searches of flu, headache, sick, chest pain and cough as a single signal |
feels_like | A combination of temperature, wind chill and heat index values |