Selected Dataset: drug$2000.$csv

The $$drug$2000.$csv$$ dataset includes medical records of patients, with various attributes. It has six columns. The first column, $$Age$$ shows the age of the patients and contains numerical values. The second column, $$Sex$$ indicates the gender of the patients and represents categorical data. The third column $$BP$$ categorizes patients based on their blood pressure levels, also presented in categorical format. The fourth column $$Cholesterol$$ contains information on the patients cholesterol levels, which are also presented in a categorical data. The fifth column $$Na$\_$to$\_$K$$ the ratio of sodium to potassium in the patients body is expressed as numerical values. The sixth and last column $$The Drug$$ specifies the type of drug recommended for each patient and contains categorical data.

This dataset is commonly used to support medication recommendations. Machine learning models can be developed to determine the most appropriate drug type based on the patients demographic and medical characteristics. Analysing this dataset can personalise treatment plans and help healthcare professionals make more effective drug recommendations.

Selected Algorithms: Random Forest, K$-$Nearest Neighbours, Multilayer Perceptron $($MLP$)$

Random Forest: Random Forest is an ensemble learning algorithm that combines multiple decision trees. Each decision tree is trained on a random subset of the dataset during training, and at each split, a random subset of features are considered. During prediction, each tree provides an independent prediction, and the final prediction is determined by taking the majority vote $($for classification$)$ or averaging $($for regression$)$ over all trees in the forest.

K$-$Nearest Neighbours: K$-$Nearest Neighbours is simple yet powerful algorithm used for classification and regression tasks. It predicts the class $($for classification$)$ or the value $($for regression$)$ of a new data point based on the majority class or mean value of its nearest neighbours in the feature space. During prediction, KNN finds the nearest neighbours of the query point by measuring distances, typically using Euclidean distance, in the entire training dataset. Explicit training is not required.

Multilayer Perceptron $($MLP$)$: MLP is a type of neural network that consist of layers of nodes, including an input layer, one or more hidden layers, and an output layer. Each node is connected to every node in the subsequent layer, with weights associated with each connection. During training, MLP adjusts these weights using backpropagation with gradient descent to minimize a predefined loss function, such as cross$-$entropy or mean squared error. MLPs are capable of approximating complex functions, which makes them suitable for a variety of tasks, such as pattern recognition and function approximation.