Output Format¶

The pipeline generates prediction files in the results/ directory.

Output Files¶

Two types of prediction files are generated:

Output Files¶
File Pattern	Description
`Predictions_<MonthRange>_District_<Date>.csv`	District-level predictions with threshold analysis
`Predictions_<MonthRange>_<State>_<Date>.csv`	State-level aggregated predictions

Example filenames:

District Prediction Columns¶
Column Name	Data Type	Description
`district`	String	District identifier (e.g., “district_374”)
`recordDate`	Date	Date of the prediction record (YYYY-MM-DD)
`ISOWeek`	Integer	ISO week number of the year
`thresholdMethod`	String	Threshold calculation method: “historical” or “previousNweeks”
`Mean`	Float	Mean of historical cases for threshold calculation
`StdDev`	Float	Standard deviation of historical cases
`Zero`	Float	Lower bound (typically 0.0)
`Inf`	String	Upper bound indicator
`T0.00`	Float	Threshold tier 0 (baseline)
`T1.00`	Float	Threshold tier 1 (elevated)
`T2.00`	Float	Threshold tier 2 (high)
`startDatePredictedWeek`	Date	Start date of the week being predicted
`dateOfComputingPrediction`	Date	Date when the prediction was computed
`regionID`	String	Region identifier matching the district
`prediction`	Float	Predicted number of dengue cases
`model`	String	Model used (typically “ensembleModel”)
`predictionZone`	Float	Zone classification based on threshold tiers

State Prediction Columns¶
Column Name	Data Type	Description
`dateOfComputingPrediction`	Date	Date when prediction was computed
`startDatePredictedWeek`	Date	Start date of predicted week
`regionID`	String	State identifier (e.g., “state_29”)
`prediction`	Float	Predicted number of dengue cases for the state
`thresholdMethod`	String	Threshold calculation method used
`predictionZone`	Integer	Alert zone classification (0=low, higher=elevated risk)
`model`	String	Model used for prediction

Temporal Resolution: Weekly predictions
Prediction Horizon: Multiple weeks ahead (typically 2-4 weeks)
Threshold Methods: Two methods for alert classification:
- historical: Based on historical mean and standard deviation
- previousNweeks: Based on recent N weeks of data
Prediction Zones: Numerical classification indicating risk levels
Model Type: Ensemble model combining multiple forecasting approaches

District-level:

district_374,2025-10-27,42,historical,4.5,0.707,0.0,inf,4.5,5.207,5.914,2025-10-27,2025-10-27,district_374,0.819,ensembleModel,1.0

State-level:

2025-10-27,2025-10-27,state_29,21.49,historical,0,ensembleModel