Have a question about this requirement?

Thank you! Your submission has been received!
Oops! Something went wrong while submitting the form.
Requirement

Multiple predictive algorithms are available to use for any account in the forecast

Functional Area

Planning

Industries
All
DETAILS

Description

This requirement enables the user to choose the most suitable predictive methods according to their specific needs, enhancing both the accuracy and efficiency of the predictions made. This implies that the system offers predictive analytics and includes more than one methodology.

Example Use Case

Scenario: An international pharmaceutical company uses CPM software for forecasting unit demand. They need to utilize different predictive algorithms for different segments of their business, considering each segment has varied market dynamics and business models.

Solution: With access to multiple predictive algorithms within their CPM software, they can adjust their strategy for each segment. They may use a seasonal trend algorithm for their flu vaccine production, given the seasonality of flu prevalence. Meanwhile, for a newly launched product in a less mature market, a different, more experimental predictive model may be chosen to forecast growth.

Considerations

While a product might have many predictive methods to choose from, it is important to determine if they cover multiple categories. For example:

Time Series Analysis Moving averages, Exponential smoothing, Autoregressive Integrated Moving Average (ARIMA) models
Regression Analysis Linear Regression, Multiple Regression
Machine Learning Model Decision trees, Random forest, Gradient Boosting

In the table above, you'll notice Time Series Analysis which includes three different types. If this is all you need, great. Many systems on the market offer Time Series, which is the most common. If you have a need to get more sophisticated, investigate if there are other methodologies available.

Questions to Ask a Vendor

  • Algorithm variety: What range of predictive algorithms does your system provide? Do you offer more than just Time Series Analysis?
  • Applying algorithms: How user-friendly is the interface for selecting and applying these algorithms?
  • Support: How much technical support is required to effectively use these predictive algorithms?
  • Machine learning: Do you use any aggregate data to improve your machine learning capabilities?