Why You Use Spend Classification
Use Spend Classification as a tool to review any of your existing Oracle Fusion Cloud (requisition, purchase order, invoice, expenses) spend data and either refine the accuracy of its current categorization or assign a category for unclassified transactions.
Spend Classification requires a data administrator to train the tool to generate category predictions using a batch process that can be repeated as many times as needed to yield satisfactory results. To help reduce the time invested to prepare training data, you can use an automatic sampling algorithm that analyzes existing spend records for patterns and uses this to generate training set data that can be reviewed for accuracy.
Reclassification of spend information is carried out using a batch process that can take multiple hours to execute when dealing with large volumes. After each batch is processed, your administrator can review the results, make manual corrections, and then assess what improvements can be made to the learning process. These can be revisions to the underlying training metadata, any manual updates to improve the classification model, or enhancements to the machine learning model using incremental training set data from the most recent changes in spend patterns.
Spend Classification also lets you build multiple business rules to normalize spend data before the classification process or to automate the classification correction process on the basis of key attributes such as supplier, item, or description. After you're satisfied with the classification results for a given batch, you can set it to approved. The approved reclassification details will then be used in the analytics dashboards to show new and correctly classified spend data, thereby providing useful spend information for your organization.