Configuring and Training CIC Advisor Models with Aconex Data

A model can be selected, trained or retrained for each predefined problem identified as an area of improvement in Aconex.

Note: In the CIC Advisor application, for an Aconex project, users can indicate if a risk has been identified by indicating Yes or No in the User Feedback field. You can choose to retrain the model based on the Create Date To retrain a model,

To train a model with Aconex data:

Sign in to the administration application of CIC Advisor.
http://<host>:<port>/cicadmin
In the sidebar, click ML Workbench.
From the Data Source list, select an Aconex data source.
From the Problem list, select any of the following predefined areas that you want to focus in CIC Advisor:
- Change Request
- Field Change Request
- Field Health and Safety
- Health and Safety
- Litigation
- Review Delay Prediction
In the Model section, select the Current Model check box to indicate the current model to be used for training purposes and then click Save.
The following information additionally displays for each model:
- Model Name: A user-friendly name for a CIC Advisor model.
- Current Model: Select the check box to indicate the model being trained currently.
- Accuracy: The ratio of correct predictions to the total number of predictions.
- Precision: The frequency of being correct with a positive prediction, and how many are actually delayed out of all the risks that are predicted to be delayed.
- Recall: The ability of a model to find all the positive results actually in the data. That is, of all the activities that are actually delayed, how many the model correctly identified.
- Total Feedback: The total number of feedbacks provided for the model until today.
- Last Run Feedback: The number of feedbacks provided after the model was last retrained. This number is treated as new data for retraining the model.
- Last Run unique feedback: Displays a count of the unique feedbacks received since the last ETL run.
- Create Date: The date on which the model was added.
To train a model:
1. Click Train Model(s).
2. In the confirmation dialog box, click OK.
  In the Training Records section, a new row displays the following information when a model is retrained:
  - Custom Name: The name of the Aconex data source
  - Etl Name: The name of the ETL process.
  - Training Start Date: The start date and time when the model was trained or retrained.
  - Training End Date: The end date and time when the training completed.
  In the Model section, a new row is displays when the model completes training or retraining.

Related Topics