Answer these two independent questions:
What do these percentages tell you about the downprojection? Are these good values in your opinion?
Wouldn't it be better then to not standardize the features? Our PCA must be really good if such a high percentage of total variance is explained. Or is there maybe more to it? Discuss the reason for this significant difference when not using standardization. What effect do we see here?
Hint: for your explanation, check the statistical summary of the features.
Compare the standardized 2D and 3D PCA plots using all 30 features. Rotate the 3D plot to at least two substantially different viewpoints. Keep the data, target and preprocessing unchanged.
Discuss briefly the differences that you can observe between the 2D and the 3D PCA:
Attach three images: the 2D plot and the same 3D projection from two different viewpoints.
Compare the standardized and unstandardized 2D PCA results. Consider both the variance percentages and the class colours.
Someone concludes: "The unstandardized representation is necessarily better for this assignment because it retains almost 100%." Which response is best justified?
Inspect the standardized t-SNE plot with perplexity 30 and compare it with the standardized PCA plot. The app uses the diagnosis labels for colouring after calculating each projection.
Which statement about this comparison is justified?
State the exact printed output and describe what the program calculates. [2 marks]
A student changes the loop to:
while current <= 7:
if current % 2 == 0:
continue
total = total + current
current = current + 1
The initial values are still current = 1 and total = 0.
Explain why this version never reaches the print statement. State the values of current and total once the program becomes stuck. [2 marks]