Model Generation
Monetai's AI model automatically generates buyer prediction models based on user behavior data collected through the SDK.
Model Generation Process
1. Data Collection Phase
When the SDK is installed and events are collected, the following process occurs:
- User Behavior Data Collection: Based on user log data collected by clients
- Purchase Event Collection: Events such as free trial starts, paid subscription conversions, renewals, etc.
- Data Accumulation: Wait until sufficient data is accumulated
2. Model Training Start Conditions
Model training automatically begins when the following conditions are met:
- ✅ 100 or more purchasers and 100 or more non-purchasers collected within 3 months of data
Model training is fully automated and requires no additional configuration or intervention. The system checks the number of purchasers and non-purchasers daily, and when 100 or more of each are collected within 3 months of data, the model automatically begins training.
Model Accuracy and Status
Accuracy Definition
Model accuracy is defined as the recall value for the buyer class:
- Recall: The ratio of actual buyers correctly predicted as buyers by the model
- High Accuracy: Accurately identifies buyers to prevent unnecessary discount offers
- Low Accuracy: Incorrectly predicts buyers as non-buyers, resulting in revenue loss
Accuracy Levels and Status
Model accuracy is categorized into the following levels:
Accuracy Level | Recall Value | Description |
---|---|---|
High | 0.8 or above | Very accurate prediction of buyers |
Medium | 0.6 ~ 0.8 | Appropriate level of prediction accuracy |
Low | Below 0.6 | Prediction accuracy improvement needed |
Optimizing | - | Purchase data not yet accumulated, or model being generated based on purchase data |
It is recommended to start campaigns when accuracy is medium (0.6~0.8) or above. At this point, the model has secured sufficient prediction accuracy to conduct effective promotion campaigns.
Real-time Prediction Accuracy Check
The model's training status and accuracy can be checked through "Real-time Prediction Accuracy" in the right sidebar of the Monetai dashboard.
Model Retraining
Model retraining functionality is currently not supported. If you need model retraining, please contact support@monetai.io separately.
Next Steps
Once the model is generated, proceed to the next step:
- Start Campaign: Begin your first promotion campaign
Support
If you encounter any issues during the model generation process, please contact support@monetai.io.