Exam DP-100 topic 6 question 6 discussion

Is there anyone could explain why we should use distance error metric not classification error metric (e.g. F-score, etc) as last step? Given this is classification not clustering problem

Therefore, the sequence might look like this: a- Define a cross-entropy function activation: This function is more suitable for multi-class classification problems. It quantifies the difference between two probability distributions: the predictions made by the model and the actual distribution. b- Add cost functions for each target state: By adding cost functions for each class, the model can optimize its predictions for each category. c- Evaluate the classification error metric: After the model has been trained, evaluate its performance using a classification error metric. This metric will tell you how often the model's predictions are incorrect. So, the sequence would be A -> B -> C

C) Evaluate the classification error metric: This is the first step to evaluate the performance of a classification model, such as the crowd sentiment models. The classification error metric measures the proportion of misclassified samples in the dataset and provides a general idea of how well the model is performing. B) Add cost functions for each target state: Cost functions are used to penalize the model for incorrect predictions and to optimize the model's parameters during training. In the case of the crowd sentiment models, adding cost functions for each target state would enable the model to learn the differences between them and adjust its predictions accordingly. F) Define a sigmoid loss function activation: The sigmoid loss function is a popular choice for binary classification problems such as sentiment analysis. It returns a value between 0 and 1, which can be interpreted as the probability of a sample belonging to a particular class. Defining this function as the activation function for the output layer of the model would help to improve its performance.

Sentiment analysis is a multi-class classification, so relevant answer for how to evaluate result is 1, 2, 3 that is 100% sure, what is the correct order??? I do not know ...

Cross entropy is a loss function not an activation function but also, in my opinion, defining a cost function for all target states is not correct (shouldn't the cost function be one?); I think at most we can define a cost function for each component (clustering, classification) and evaluate distance metric and classification metric.

ans: 1,2,3

sorry 3,2,4 boxes

3,2,1 boxes

should be 3rd option.