Hello,
I am new to NLP. I’ve started using rasa nlu for my dataset of 35k examples. I’ve trained using tensorflow_embedding. I got the below results.
2019-05-08 07:56:15 INFO rasa_nlu.model - Finished training component.
2019-05-08 08:01:51 INFO __main__ - CV evaluation (n=10)
2019-05-08 08:01:51 INFO __main__ - Intent evaluation results
2019-05-08 08:01:51 INFO __main__ - train Accuracy: 0.952 (0.003)
2019-05-08 08:01:51 INFO __main__ - train F1-score: 0.947 (0.003)
2019-05-08 08:01:51 INFO __main__ - train Precision: 0.954 (0.003)
2019-05-08 08:01:51 INFO __main__ - test Accuracy: 0.932 (0.005)
2019-05-08 08:01:51 INFO __main__ - test F1-score: 0.927 (0.006)
2019-05-08 08:01:51 INFO __main__ - test Precision: 0.932 (0.006)
2019-05-08 08:01:51 INFO __main__ - Entity evaluation results
2019-05-08 08:01:51 INFO __main__ - Entity extractor: ner_crf
2019-05-08 08:01:51 INFO __main__ - train Accuracy: 0.985 (0.000)
2019-05-08 08:01:51 INFO __main__ - train F1-score: 0.985 (0.000)
2019-05-08 08:01:51 INFO __main__ - train Precision: 0.985 (0.000)
2019-05-08 08:01:51 INFO __main__ - Entity extractor: ner_crf
2019-05-08 08:01:51 INFO __main__ - test Accuracy: 0.983 (0.001)
2019-05-08 08:01:51 INFO __main__ - test F1-score: 0.982 (0.001)
2019-05-08 08:01:51 INFO __main__ - test Precision: 0.982 (0.001)
2019-05-08 08:01:51 INFO __main__ - Finished evaluation
Is there any way to check where is my dataset lacking or incorrect prediction while the dataset is being tested ?