Problem
MindsDB offers the ability to create or train models on your data. When we mention machine learning models and training, we must consider the loading and processing that often takes a long time. In MindsDB, users lack a good indicator to check their model status and must rely on SQL queries for this purpose.
Opportunity
We have the opportunity to improve the model creation process by incorporating a real-time, intuitive indicator of model status. This will provide instant feedback on whether model training is completed or not, thereby reducing the need for SQL queries.Details
Users can track the model creation process from the tree menu. We send an email when the model finishes training, and users can view all processes from the notifications icon in the top-left corner.
There are multiple scenarios for the model process that users may encounter.
Case 1 User on Mindsdb SQL Editor
We notify the user when the model starts the training process, showing the progress and completion status. Users can track all previous or old trained models from the notifications.
To make it more visible and noticeable to the user, we used a bouncing ball animation. When the user clicks on "Run", we instantiate a small squishy model icon that bounces and flies up, landing where the models training list is located.
Because users might close the sidebar, we need to display a small bar when the user clicks "Run", This bar indicates the number of models currently training and can be hidden.
Additionally, users can clear completed training models from the list.
Model status bar and icons for use in training list and notifications
Case 2 The user has the MindsDB tab open but is exploring another app
We notify the user about the model status using browser notifications.
Case 3 User started training model and closed MindsDB web app or machine We send an email when the model training is completed.
Users can manage all notifications and email preferences from their account settings. They can also adjust animation settings to reduce visual effects.
