forecast

Invoked by: vaxstats forecast

The forecast command performs forecasting on prepared data using a specified statistical forecasting model. It allows you to apply various forecasting algorithms to your time series data and generate predictions.

Usage

vaxstats forecast <file_path> <sf_model> [options]

Examples

vaxstats forecast data.csv statsforecast.models.ARIMA --baseline_hours 48

vaxstats forecast prepared_data.csv statsforecast.models.ARIMA \
  --baseline_days 3 --sf_model_kwargs "{'order': (1,1,1)}" \
  --output_path forecast_results.csv

Required Arguments

Argument	Description	Example
file_path	Path to the input data file (CSV format).	data.csv, prepared_data.csv
sf_model	The forecasting model class to be used.	statsforecast.models.ARIMA

Optional Arguments

Option	Description	Default	Example
--baseline_days	The time window in days for the training set.	N/A	3, 7.5
--baseline_hours	The time window in hours for the training set.	72.0	48, 120
--sf_model_args	Positional arguments for the forecasting model constructor.	()	"(1, 'string', [1, 2, 3])"
--sf_model_kwargs	Keyword arguments for the forecasting model constructor.	{}	"{'order': (1,1,1), 'seasonal_order': (0,1,1,12)}"
--output_path	Path to save the output DataFrame with forecasted values.	output.csv	forecast_results.csv

Note

You must specify either --baseline_days or --baseline_hours, but not both.

Input

The input file should be a CSV file containing at least the following columns:

• ds: Datetime column
• y: Target variable column, \(y\)

This file is typically the output of the prep command or a similarly structured dataset.

Output

The forecast command will generate a CSV file (default name: output.csv) with the following columns:

1. All original columns from the input file
2. y_hat: The forecasted values, \(\hat{y}\)
3. residuals: The difference between actual and forecasted values (\(y - \hat{y}\))

Forecasting Model Specification

• Use the sf_model argument to specify the full path to the forecasting model class.
• Use --sf_model_args and --sf_model_kwargs to pass additional arguments to the model constructor.
• Arguments should be specified using Python literal syntax.

Examples of Model Specifications

1. ARIMA model with specific parameters:

   vaxstats forecast data.csv statsforecast.models.ARIMA \
       --sf_model_kwargs "{'order': (1,1,1), 'seasonal_order': (0,1,1,12)}"
   

2. Simple exponential smoothing:

   vaxstats forecast data.csv statsforecast.models.SimpleExponentialSmoothing
   

Tip

Experiment with different baseline windows and model parameters to find the best fit for your data. You can compare the residuals to evaluate model performance.