Page 45 - FULL REPORT 30012024

may be impacted by incomplete or poor-quality data, which might produce

inaccurate or biased findings.

According to (Berendsen, 2019), The lack of transparency in some machine

learning models presents another difficulty. These models, which are
frequently referred to as "black boxes," can be challenging to analyze and

comprehend. Due to this lack of transparency, it is difficult to spot and correct

any mistakes or biases that may have crept into the model's decision-making.

Another issue with machine learning is overfitting. A model may perform
extraordinarily well on the training data when it is overly complex, but it may

not generalize to fresh, untried data. The model's ability to provide precise
predictions or classifications in real-world situations may be jeopardized by

this overfitting problem.

According to Pappadà (2022), machine learning also raises ethical issues.

Machine learning models may perpetuate biases found in the training data or
biases built into the algorithms, resulting in discriminating or unfair results.

It is a continuous challenge to address these moral questions and guarantee
fairness in machine learning algorithms.

Furthermore, human monitoring is limited as machine learning models get

more complicated. It can be difficult for people to comprehend how complex

models make decisions, which raises questions about possible unintended
consequences or mistakes that might go undetected.

2.6 Machine Learning for Prediction

In many fields, machine learning is a potent tool for prediction. On top of

that, machine learning has been used to answer numerous clinical issues in
electronic health data, where it may be particularly effective at spotting novel

traits or nonlinear correlations. Machine learning is a vast field that employs

optimization techniques to create computers that automatically identify
patterns in data and forecast desired future results.

28