april 2025
AI Platform V 2.33
✨ Optimized predictions from and to all integrations
Now you can manage both single-time and recurring predictions in a single flow—just choose the mode!
Select whether to run a single or recurring prediction without extra steps.
Recurring predictions now work seamlessly with multiple tables.
In addition to the standard .csv download, you can now export data to multiple systems.
Export Predicted Data with Flexibility: Append or Update
📊 New scatter plot dashboard in value models
Now you can maximize the predicted value with advanced analysis tools:
Visualize data distribution and identify trends at a glance in the scatter plot.
Accurately assess the model’s error range against real values.
Monitor the quality and confidence of predictions in real time.
📢 Filtering and replacement management
Our system now automatically saves the original dataset before applying any filters. You can always revert to the original dataset without losing information or needing to re-upload it.
🔍 Suggested Filters
The system now automatically suggests the most relevant filters for your datasets, creating a consistent and optimized structure for analysis.
Applied filters are also recommended for other datasets, regardless of when they were uploaded.
Avoid setting the same filters multiple times.
Maintain uniform filtering across all your data.
🎲 Random sampling (Advanced mode)
The system now allows you to import only a random portion of the data instead of the entire dataset. This can help speed up processing and reduce the data load.
📈 Control group in Best Option models
- It allows you to monitor the model’s performance on a specific subset of data, ensuring more accurate analysis.
- It helps validate results and optimize predictions in complex scenarios.
🧠 New processing stages in the "Advanced" section of the model
It allows for further control over the customization of the model.
⭐️ Selection of the primary table for star join connections
Now you can easily select the main table in multi-table flows, without needing to load it first as before.
🪟Time windows: customizable values
Now you can enter custom time intervals, allowing for greater flexibility in both predicting the likelihood of a positive outcome for a subject and performing general analysis time.