Iris Dataset Project

Introduction

The Iris dataset is a classic dataset used in machine learning and statistics. It contains 150 samples of iris flowers, with three different classes: Setosa, Versicolor, and Virginica. Each sample has four features: sepal length, sepal width, petal length, and petal width.

Data Visualization

• Histograms and Bar Plots: Plotted histograms for numerical data and bar plots for categorical data.
• Pairplot: Visualized pairwise relationships between features using a pairplot.
• Covariance Matrix and Correlation Coefficient Matrix: Generated and visualized the covariance matrix and correlation coefficients between columns.

Outlier Detection and Removal

• Boxplot Analysis: Identified outliers using boxplots.
• IQR Method: Removed outliers using the Interquartile Range (IQR) method.

Data Preparation

• Data Splitting: Split the dataset into training (60%), validation (20%), and testing (20%) sets.
• Normalization: Normalized the data using z-score normalization with StandardScaler.

Model Evaluation Metrics

Used the following metrics to evaluate models on both training and validation sets:

• Accuracy Score
• F1 Score
• Precision Score
• Recall Score

Classification Models

Implemented the following models without GridSearch:

• GaussianNB
• RandomForestClassifier
• DecisionTreeClassifier

GridSearch Implementation

Performed GridSearch for hyperparameter tuning on the three models without using cross-validation, as the validation set was already separated.

Dimensionality Reduction

Applied Principal Component Analysis (PCA) for dimensionality reduction, keeping 95% of the variance. After applying PCA, tested the three models again with and without GridSearch.

Best Model Selection

Selected the best model based on the highest F1 Score on the validation data. The best model, random_forest_pca_grid, was then tested on the test data, achieving a final F1 Score of 0.93.

If you like this project, please give it a ⭐ star.