A newly created classifier does not have information, so we have to train it. To do so, we have to provide a training set.
For now, we use pre-classified texts (alternative A in "Your first classifier"). Pre-classified texts must be provided as excel files and must have two columns.
The first column needs to contain the class/category name, while the second one is the open feedback text.
In "Your first classifier", we had a simple sentiment analysis with only two categories: "Negative" and "Positive".
You can find the example file, MyFirstClassifier - Training.xlsx, attached to this article right at the bottom of it.
Note that if you have more categories the process is the same.
The category names are case sensitive.
Next, click the "Upload data" in the top VoC Hub menu and you'll be able to proceed with uploading on the opened page.
Choose the classifier set you want to apply the training to and select the "training pre-classified" type. You can move your input file via drag and drop on the field with disk symbol or you click on it and choose the file from your computer. You can also provide a title for future reference.
Now click on the "Submit" button and the training set will be uploaded.
You can track the progress as well as also see all the details of your uploads by going to the "List uploads" section:
Once the training has completed you can see statistics for it in the "statistics" section:
Example training file: