In FreshEngage version 4.0, subscribers can set up their own Machine Learning model to automatically tag, classify, categorize and assign freshdesk tickets . He/She can use either his/her own ticket data in the form of CSV or He/She can use existing tickets from FreshDesk as training data to create the model. In order to train model using existing tickets, Customer need to have at least 100 tickets in his/her freshdesk account. 

Log into https://freshenage.iengage.io the freshengage admin panel  follow below instruction to train and activate your own model..

Steps for training Model :-

  1. Login into https://freshenage.iengage.io .

  1. Go to the Training page by clicking “Training” from the navbar.

  2. Choose your way for training. 

  3. Use existing data
    1. Click on “Import” button.

    2. Choose your CSV file. File type CSV is mandatory.

    3. Give some name to model.

    4. Click on the Train button.

    5. Pre Training Analysis. Refer to the next bold point.

    6. After successful training, You will see your trained model record in the table. If you want to use this model for ticket classification. Hit the radio button in front of your model name to set as the default model.

    7. To test the accuracy of your trained model, you will get the option to test in the same row of that record. Where you need to provide some text and the system will return confidence in float.

  4. Use existing tickets.

    1. Click on the “Train Model Using Freshdesk Tickets” button.

    2. Choose  “Since Date” wherefrom which date you want to fetch tickets. The date format should be YYYY-MM-DD.  The date is mandatory.

    3. Select the source that you want to fetch the tickets for model training.

    4. Click on the Train button. 

    5. Pre Training Analysis. Refer to the next bold point.

    6. After successful training, You will see your trained model record in the table. If you want to use this model for ticket classification. Hit the radio button in front of your model name to set as the default model.

    7. To test the accuracy of your trained model, you will get the option to test in the same row of that record. Where you need to provide some text and the system will return confidence in float.

Pre Training Analysis

While training models from both ways you will get pre-training analysis. This is an analysis of data that you have provided for model training. If all data is sufficient and valid then you will get the option to train the model. In other cases, you can correct your data or cancel the training for that particular model.



Add Required Ticket Types

In order to classify ticket we need to add some custom types in ticket fields. These types will be provided after ML model training. If your model has trained using existing freshdesk tickets then no need to add types in Freshdesk. In other case Its mandatory.

To add new types your Freshdesk, go to Admin -> General Settings -> Ticket Fields and create new custom types ignore if you already have.