Discussions

Hi,

The objective is to forecast future sales of high-value, high-selling medicines using medicine sales dataset that spans one year. However, some days in the dataset have no sales, and these dates are filled with zeroes. Despite implementing an advanced forecasting model (TimeGPT) and adding additional features like holidays, weekends, and cyclic seasonality, the accuracy of the predictions is still not meeting expectations.

I’m seeking suggestions on adjustments or additional techniques that could help improve the accuracy of these forecasts.Is there any other methods to enhance the model’s performance for this type of time series data?