DashboardSettings
Settings
Configure prediction model defaults and data processing
Prediction Model Defaults
Set default model behavior for new sensors
days
7d30d60d90d
Changes apply to newly added sensors
Bulk Apply Model
Override model selection for all sensors in a specific area
Full History(uses default above)
Bulk apply will override any per-sensor model selections in the chosen area. Individual sensors can still be toggled back from the Predictions page.
Data Processing Pipeline
Auto-clean methods active for rolling model
IQR Filter
Remove values outside 1.5x interquartile range
Z-Score Filter
Remove data points with |z| > 3 standard deviations
Frozen Signal Removal
Detect and remove constant-value windows (std dev = 0)
Sensor Quality Flag Filter
Exclude points with quality_flag > 0
Rate-of-Change Filter
Remove physically impossible value jumps