How to Optimize Accuracy Rate?

After creating a custom machine learning model at Kimola Cognitive, you will be able to see its accuracy rate. Accuracy Rate is essential for ML models since you need your data to be analyzed correctly. We recommend having a minimum of 85% accuracy rate for your custom machine learning models.

A step-by-step guide to optimizing your custom machine learning model (See the video below!) 

• First, sign in to Kimola Cognitive.
• On the dashboard, find Models on the left bar. 
• Find and click on the custom machine learning model that you’d like to optimize the accuracy rate.
• To see the accuracy rate of a custom ML model, click Overview on the left menu, and you will be able to see the accuracy rate of the model.
  
  

• Go to Explore on the left menu to check your training set is labeled correctly. You can change the labels on this page.
• Remember to have a minimum of 500 content for each label (2,500 is suggested). If your accuracy rate doesn't meet your expectations, try adding more content to each label.
• Remember to have enough content for each label and the content is distributed equally on each label.
• Keep character counts similar.
• Remember to have a minimum of 2 and a maximum of 8 labels.
• Click on the “New Record” button to add new labeled content on the right side.
• You can also download your labelled content as .xls format by clicking the Download button on the right corner and work on it offline.

See dashboard in action: