Analyze Accounting Transactions

Form ID: (GL510000)

Note: This form appears in the system if the GL Anomaly Detection feature is enabled on the Enable/Disable Features (CS100000) form.

On this form, you can perform initial training of the machine learning model, analyze transactions, and get the results of transaction analysis.

The form provides information about only companies that have the Use for Anomaly Detection check box selected on the Companies (CS101500) form. For each company included in processing, the system detects the financial periods from which it is possible to select posted transactions. Then the system sends the transactions to the cloud service. After training or analysis has been performed, the system loads the response from the service and updates the data in the table.

The form is visible to only users that are assigned to the Administrator, GL Admin, or AcumaticaSupport role on the User Roles (SM201005) form.

Tip: You can create an automation schedule for each process that can be run on this form.

Form Toolbar

The form toolbar includes standard and form-specific buttons and commands. For the list of standard buttons, see Form Toolbar and More Menu. The form-specific commands are listed in the following table.


Button	Description
Process	Initiates the action specified in the Action box for the selected companies.
Process All	Initiates the action specified in the Action box for all companies listed in the table.

Selection Area

In this area, you select an action to be performed in the Action box.


Element	Description
Action	The action that the system will perform when you click Process or Process All on the form toolbar. The following options are available: Train Model: When the option is selected, the system loads active companies for which the Use for Anomaly Detection check box is selected on the Companies (CS101500) form. To use this action, the service status of each company should be Pending Training, Training Failed, or Ready for Analysis. The system then sends the GL transactions of each company to the cloud service by using an API. This action is performed for initial training of the model based on all companies after a system upgrade, for initial training based on a new company, or for training based on existing companies. Initial training is performed based on 18 previous calendar months and an anomaly score is assigned to each group of parameters. Analyze Transactions: When the model has been trained and you select this option, the system loads active companies for which the Use for Anomaly Detection check box is selected on the Companies form. To use this action, the service status of each company should be Ready for Analysis. The system then sends GL transactions from the first open period of each company to the cloud service by using an API. You can send transactions for analysis as many times as needed. In this case, only newly posted transactions will be sent to the cloud service. Get Status: When you begin processing for all companies or only selected ones and select this option, the system loads all companies with the service status of Training in Progress and Analysis in Progress. For companies with the Training in Progress status, when training is completed successfully, you can send transactions from an open period for analysis. For companies with the Analysis in Progress status, when the results of transaction analysis are ready, the cloud service returns the anomaly scores to the system. You can view the anomaly scores on the Review Anomaly Predictions (GL406000) form by selecting the needed selection criteria.

Table 1. Table Columns
Column	Description
Included	An unlabeled check box that you select to include this company in processing when you click Process on the form toolbar.
Company ID	The company ID of a company that is active and has the Use for Anomaly Detection check box selected on the Companies form.
Company Name	The company name of the company.
Service Status	The current status of the GL anomaly detection service for each selected company. The following statuses are available. Pending Training: Initial training has not been performed for the company. This status is the default one after an upgrade and for new companies created after the upgrade. Training in Progress: Data has been sent by the training process, and the model is not yet ready. Training Failed: The training failed because the model could not perform training on the data that was sent to it. Ready for Analysis: The model is ready for transaction analysis and prediction. Analysis in Progress: Data has been sent for analysis, and the model has not yet returned confidence scores. A confidence score is an indicator that the transaction is likely an erroneous one. The higher the confidence score, the more likely the transaction is to be erroneous.
From Period	This column is filled in by the system depending on the action selected in the Action box and the service status of the company. The value in this column is calculated by the system based on the To Period value. If Train Model is selected and the service status is Pending Training, the system fills in the period corresponding to the 18th month in the past from the To Period setting. If Train Model is selected and the service status is Training Failed or Ready for Analysis, the system fills in the earliest unsent period. If Analyze Transactions is selected and the service status is Ready for Analysis, the system fills in the first open period. You can select another period. An empty value in this column means that no transactions will be sent if this company is selected for processing, and the system displays a warning message.
To Period	This column is filled in by the system depending on the action selected in the Action box and the service status of the company. If Analyze Transactions is selected and the service status is Ready for Analysis, the system fills in the period previous to the period of the business date. You can select another period. An empty value in this column means that no transactions will be sent if this company is selected for processing, and the system displays a warning message.