Business Assist Forecast API Concepts
Historical Data
Historical volume data is data collected during normal business operations and is used to generate forecasts of future volume.
This historical data should contain date information, paired with daily volume metrics in a simple format. This input can be collected in JSON format or a CSV file.
Related Data
Special events and holidays can affect volumes in ways that can skew the results of weekly or monthly averages. These outliers must be accounted for to accurately forecast case volumes. Business Assist Forecast API adjusts its forecast models based on dates you provide.
Business Assist Forecast API lets you set the seasonality to analyze case volume data on a hourly, daily, weekly, monthly, quarterly or yearly time frame.
Data Models
The Forecast API uses Frequency and Bayesian statistical methods to calculate the predictive distributions to analyze the historical data and produce a confident forecast of future volume. The models can make predictions for various time periods, such as hourly, daily, weekly, monthly, quarterly or yearly.
Forecast parameters
The Forecast API can forecast for a date range up to six months, depending on how many days of historical data are available and used. In general, the models can forecast half of the input date range, with the following conditions:
- If the historical data time range is less than 12 months, forecasting time range is the half of the input time range.
- For example, eight months of historical date range can forecast for next four months.
- If the historical range is at least 12 months (up to 24 months), the report will forecast for next six months.
The historical data must meet the following minimum requirements for the models to generate forecasting.
- At least two weeks of historical data is available.
- For case and conversation forecasting, the number of days when no cases is created should be less than 30 percent per product.
- To achieve better forecasting result, the average daily volume of greater than 50 is recommended.
First time user of Business Assist Forecast API?
Find resources for getting the most out of the Business Assist Forecast APIs:
- Overview: High-level introduction to Business Assist Forecast API.
- How to use: Learn how to create your first forecast forecasting predictor with the Forecasting API.
- API reference: Familiarize yourself with the Business Assist Forecast API actions and data types.