The annotation statistics gives you an overview over of the distribution of already annotated product distributionproducts.
On the top, the total number und and the number of already annotated product products are displayed. Especially the The second number and its percentage is are especially helpful during the annotation because you get an idea how many products are annotated yet. The annotation, as the prediction model needs a certain number amount of products to achieve a good prediction for to be able to learn the most likely attribute value of unknown (not annotated) products.
Below these numbers the statistics show the , you can also see the distribution of the annotation annotations over the attribute values. It displays the total number of annotations (e.g. female was annotated 3 times) and the percentage in brackets.
...
| title | Note |
|---|
...
is especially helpful for identifying underrepresented values and consider grouping these together or even removing them, since the model would likely not be able to learn any useful patterns from just a few products.