Creating a predictive model
In this part of the guide, we will show you how to create a predictive model, step by step:
The first step is to go to the ModelLab section of Predictto.
There, select the type of analysis you want to perform in the expandable menu at the upper left. Remember to use the information in the first part of this guide as a reference so that you know which analysis is best for your needs.
Then, if you choose, you can select a range of dates so that your model only considers the data obtained within that range.
In the selection tree you will find all the devices, types of devices, and parts of devices that are registered on our platform. It’s important to note that if you chose a type of analysis that uses information from meters (like the calculation of deterioration models), the selection tree will only show the devices that are connected to one or more meters.
Once you have selected the filter parameters indicated in steps 3 and 4, click on the “Filter” button.
Now select the source of the data that will feed your model. If you chose a deterioration model analysis in step 2, in the table on the right of the screen you will see all of the meters connected to the devices and parts selected in step 4. Check the “select” box of the meter you want to feed your model. Remember that if you chose a range of dates in step 3, the model will only consider the measurements carried out within that range.
Finally, give a name to the data group that you chose to feed your model. It’s important to use a descriptive name because you will need it later to identify the model and view the diagnostics derived from it. Once you have chosen the name, click “Save and Analyze.”
By following the steps above, you will have created a predictive model. Depending on the amount of data used to feed the model, the calculation time may vary. Our algorithms are optimized to reduce this time to the minimum and we are constantly working to improve on this.
Once the model is calculated, you will find it in the Devices section, where you will see a list of all the existing models associated with your account, including the one you’ve just created. By clicking on the name you gave to the data group used to feed your model, you can view the results and diagnostics derived from it.
It’s possible to select various meters at the same time, which will generate a predictive model for each one.
This way you can create more models in fewer steps and in less time. To make multiple selections, use the “ctrl” button (“command” in macOS) and select more than one asset.
Predictive model views
Models view (view by defect): This presents a list of all the predictive models calculated and stored in your account, including information such as the name assigned to the training data group, the date the model was created, the type of analysis used, etc.
View organized by assets: Groups the models according to the assets that are registered in your Predictto account. It shows the model with the highest criticality forecast for each one.
View organized by parts: Similar to the view organized by assets, but with models grouped according to the parts of assets registered in Predictto.
Rapid access indicators
At the right side of the list, you will see an indicator in the form of a circle that indicates the critical state of an asset; green indicates that there is still time before maintenance is needed and red indicates that there’s no time left or that the optimal moment to carry out maintenance has passed.
It’s also important to mention that the rapid indicators are based on the information contained in diagnostics.
Deletion and recalculation of various models
You can select different models to delete or recalculate with just one click. If you click on the Retrain option, the analysis parameters that were first established will remain the same, but it will include the most recently collected data from the meter that was selected to create the model.
To get more information about the analysis results, just click on the “View diagnostic” button.
If you want to know more about the tools located there, go to the next section about diagnostic tools.