Check the confusion matrix: how many actual cancer cases did the model miss?

Check the feature importance: what is the most important feature? Select 'Permutation', normalize importances, use 10 repeats and score by accuracy.

Attach the plots of the ROC curve and the Confusion Matrix here. Are there any misclassifications? If yes, check the probabilies of the misclassified samples, was it a close decision or did the model clearly vote for the wrong class?

What is the area under the ROC curve?

Apply a k-NN classifier on the Breast cancer data. Check the hyperparameter optimization tab to find a good value for k. What is the lowest value for k where we still get good accuracy on the training and validation set?

Plot the mystery function and try polynomials of degrees 1, 2, 3, 5, 13, 19 and 26. Which which of these degrees fits the data points best (without overfitting)?

Create a GMM model with the estimated number of clusters. Create a contour plot that also shows the component ellipses and attach the plot here. Now check out the clustered samples in the 'Sample from Clusters' tab. Can you find out what the common topic of the posts in each cluster is? Give a list of the topics/hashtags here.

Apply a GMM model on the BlueSky data. Check the AIC/BIC Values - what is the estimated number of clusters in this dataset?