$1.$ Introduction

The primary objective of this report is to analyze a given dataset and construct predictive models to classify the data accurately. The dataset comprises various features, each contributing to the classification task. We aim to identify data quality issues, preprocess the data, and develop two machine learning models: Decision Tree and Random Forest. This report will detail each step, providing justifications and methodologies used to ensure robust model performance.

$2.$ Analytical Base Table Characterization

$2.$ Analytical Base Table Characterization

The dataset consists of $13,611$ instances and $20$ features. Each feature plays a significant role in classification. The table below characterizes each feature in the dataset:

Feature Description Type Missing Values Unique Values

Class Target variable Categorical $07$

Area Area of the object Numerical $013421$

Perimeter Perimeter of the object Numerical $013421$

MajorAxisLength Major axis length of the object Numerical $013421$

MinorAxisLength Minor axis length of the object Numerical $013421$

AspectRation Aspect ratio of the object Numerical $013421$

Eccentricity Eccentricity of the object Numerical $013421$

ConvexArea Convex area of the object Numerical $013421$

Constantness Constantness of the object Categorical $02$

EquivDiameter Equivalent diameter of the object Numerical $013421$

Colour Colour of the object Categorical $07$

Extent Extent of the object Numerical $013421$

Solidity Solidity of the object Numerical $013421$

Each feature's summary statistics and data types will provide insights into the dataset's structure and inform the preprocessing steps.

$3.$ Identifying Data Quality Issues

$3.$ Identifying Data Quality Issues

In exploring the dataset, several data quality issues have been identified:

Missing Values: There are $29$ missing values spread across various features. Missing data can lead to biased models and inaccurate predictions.

Outliers: The presence of $1287$ outliers was detected, which can significantly skew the results and degrade model performance.

Inconsistent Data Types: Some features may have inconsistent data types that need to be standardized.

Duplicate Records: No duplicate records were identified in the dataset.

Imbalanced Classes: The class distribution appears to be imbalanced, which can affect the model's ability to generalize well.

Justifications for Identified Issues:

Missing Values: Missing data can lead to incomplete analysis and poor model performance. Imputing these values is crucial.

Outliers: Outliers can distort statistical measures and lead to biased models. Addressing them ensures more robust model performance.

Inconsistent Data Types: Consistent data types are necessary for accurate data analysis and preprocessing.

Imbalanced Classes: Imbalanced datasets can cause models to be biased towards the majority class, leading to poor performance on minority classes.

$4.$ Machine Learning Approaches

$4.$ Machine Learning Approaches

Two machine learning models were selected for this problem: Decision Tree and Random Forest.

Decision Tree:

Justification: Decision Trees are easy to understand and interpret. They can handle both numerical and categorical data and do not require data scaling. They are robust to outliers and can model complex relationships.

Overview: A Decision Tree model splits the data into branches to predict the target variable. It uses a tree$-$like structure to represent decisions and their possible consequences.

Random Forest:

Justification: Random Forest is an ensemble method that combines multiple decision trees to improve classification accuracy and control over$-$fitting. It is robust to outliers and can handle high$-$dimensional data well.

Overview: A Random Forest model builds multiple decision trees and merges them to get a more accurate and stable prediction. It leverages the wisdom of the crowd by averaging the predictions of individual trees.

$5.$ Data Preprocessing Steps

$5.$ Data Preprocessing Steps

To ensure optimal performance of the machine learning models, several preprocessing steps were implemented:

Decision Tree:

Handling Missing Values: Missing values were imputed using the median of the respective columns. This method is robust to outliers and ensures that the central tendency is maintained.

Encoding Categorical Features: The Colour feature was label encoded to convert categorical values into numerical labels.

Outlier Treatment: Outliers were detected using the IQR method and handled by capping them at the $1$st and $99$th percentiles.

Feature Scaling: Scaling was not necessary for Decision Trees as they are not sensitive to the scale of the features.

Random Forest:

Handling Missing Values: Similar approach as Decision Tree, imputing missing values with the median.

Encoding Categorical Features: The Colour feature was one$-$hot encoded to create binary columns for each unique colour.

Outlier Treatment: Outliers were handled similar