Please help me understand this learning curve… Y axis have accuracy score and X axis have number of training examples
This curve tells you the trend of the performance of your model.
Generally, what happens most of the times is that our model has a lower training error, but higher validation/testing error.
This is because our model hasn’t learned to generalize well - has overfit - hasn’t learnt the underlying patterns but learnt the unwanted noise.
Your graph plots scores, which is inversely related to error( less error-> high score)
Kindly refer to this thread :
Underfitting – Validation and training error high
Overfitting – Validation error is high, training error low
Good fit – Validation error low, similar trend as the training error