R studio exam
Question 1
1)

2)
3)
4)
5)
Trace A’s primary diagonal from left to right using a square matrix (A). Multiplicities are used to count eigenvalues, and the trace of a matrix is the sum of these eigenvalues.

6)
A covariance matrix is symmetric since it relates to an array of integers.

7)
To begin, we must establish A’s eigenvalues. Bear in mind that these are the answers to the equation.

Using Laplace Expansion, this determinant may be determined and simplified. The following equation represents the result of the experiment.

By solving this equation, the eigenvalues may be determined.

8)

The rank and nullity of a matrix theorem state that if p(A)=0 if and only if A=0 since M*N Matrix a with rank r implicitly implies A is not null and r is a positive integer.
Question 2

Question 3
Question 4
a) Cancer prediction requires the use of the KNN approach, since it does not need a human-readable model. The distance between two points has an effect on prediction accuracy.

b)

Random forest ensembles, also known as random choice forests, are a frequently used technique for developing predictive models for classification and regression. There are several decision tree types that may be utilized to get the optimal response in a random forest model. This is how it works in practice, according to the Breast Cancer Wisconsin (Diagnostic Data Collection), a data collection comprised of characteristics extracted from digital pictures of biopsies.
c)
i)

ii)
iii)
iv)

Question 5
a)

b)
c)
d)
e)

Question 6
a)
In a covariance matrix, the correlations between each pair of variables are represented by a symmetric matrix, which represents the correlations between each pair of variables. The covariance matrices of multivariate data distributions in multidimensional space may be used to characterize the amplitude and direction of the distributions. It is possible to investigate how data spreads across two dimensions by altering these quantities. Eigenvalues are numbers that represent how much variation there is in the data; in this situation, eigenvalues are numbers that indicate how spread out the data is along the line, where the number of eigenvectors/values equals the number of dimensions in the data set.
b) It is standard practice in the realm of machine learning to split data into many sets for training, testing, and antivirus validation. This approach may be used to determine both the model hyper-parameter and the generalization performance. The data used to train your model is referred to as training data. It describes what your model observes and learns from the data used to train it. When we feed our model with testing data, we may anticipate that certain values would be anticipated (without seeing actual output). We evaluate it by comparing it to the actual outcomes obtained from the expected testing data. These events serve as training data, and we utilize this approach to assess the amount of knowledge gained by our model.
c) Factor analysis is used to reduce non-linear dimensions, while PCA (principle component analysis) is used to reduce linear dimensions. PCA does not make use of hyperparameters such as perplexity, learning rate, or step count. Factor analysis explicitly presupposes the existence of latent components hidden underneath the visible data. PCA, on the other hand, seeks to discover variables that are composites of those already present in our database.
Although PCA has big outliers, factor analysis is capable of addressing these abnormalities. We may utilize eigenvalues in factor analysis to decide how much variation to maintain. We can use hyperparameters to preserve distance but not variance. The significance of Principal Components Analysis cannot be overstated (PCAexploratory). In contrast to PCA, factor analysis uses the diagonals to represent the standardised error variance after all other variables have been used as predictors for that column’s (row’s) variable. This is the key distinction between the two exploratory factor analysis methodologies. Rather of using 1s or wrong variances in your variance-covariance matrix version, you may choose to use non-standardized variants (ie the variance for PCA and the unstandardized error variances for factor analysis).
In general, the factor structure and estimates are relatively comparable across the two methodologies. Factor analysis is often utilized in statistical research due to the (ostensibly) impartiality of the estimates.

d) The accuracy of knn is reliant on the data quality, which is a challenge. When a large amount of data is available, the prediction process may take a lengthy time. Additionally, it is critical to highlight that the KNN approach takes a significant amount of storage space since it requires the retention of all training data. Finally, since the whole training session is saved, it may be computationally costly. The use of KNN classifier is important because it predicts the class of a certain test observation by comparing it to the most similar observations. The effect of factors that substantially affect the distance between the observations and the KNN classifier will vary dramatically.

e) Orthogonal rotations do not allow for factor correlation and may result in the loss of critical data in scientific study, while oblique rotations allow for factor correlation and are therefore more realistic.