🚀 Spaceship Titanic Challenge – Kaggle Group Assignment

Team Name: Cosmic Learners (Group 9)
Challenge: Spaceship Titanic - Kaggle Competition
Course Assignment: Data Science Assignment 3 – Group Kaggle Challenge
Google Colab: Google Colab Link GitHub Repo: Github Link

---

👨‍🚀 Project Summary

Welcome to our repository for the Spaceship Titanic Challenge, a machine learning competition hosted on Kaggle. Our objective was to develop predictive models that determine whether a passenger was transported to an alternate dimension during a futuristic interstellar voyage.

Our group, Cosmic Learners, collaboratively built a complete and reproducible machine learning pipeline, including exploratory analysis, feature engineering, model training, hyperparameter tuning, and performance evaluation.

---

👥 Team Members

• Ayisha Fidha Maniyodan
• Diya Amith Kodappully
• Dona Uresha Pamodi Dasanayake
• Fawas Afsal
• Mohammed Nihad Kaipalli
• Sam Jacob
• Sandra Binu
• Sharon Zacharia

---

🧠 Project Workflow

This notebook showcases a comprehensive machine learning pipeline divided into the following structured sections:

1. 🧭 Exploratory Data Analysis (EDA)

• Univariate & bivariate feature exploration
• Correlation analysis
• Visualizations using seaborn, matplotlib, squarify, and pie charts
• Identification of missing values and patterns

2. 🧹 Data Preprocessing

• Missing value imputation (mean/mode strategy)
• Feature extraction from Cabin, Name, and other composite fields
• Label encoding for categorical variables
• Standardization of numerical features

3. 🧪 Model Training & Evaluation

• Logistic Regression
• Random Forest
• XGBoost
• Deep Neural Network (Keras Sequential Model)
• Evaluation metrics: Accuracy, Precision, Recall, F1 Score, AUC-ROC
• ROC Curves & Model Comparison Charts

4. 🔍 Hyperparameter Tuning

• GridSearchCV for Logistic Regression, Random Forest, and XGBoost
• EarlyStopping & ReduceLROnPlateau for DNNs
• PCA used optionally for dimensionality reduction & visualization

5. 📈 Explainability

• Model explainability with SHAP (SHapley Additive exPlanations)
• Visual insights into feature contributions for top predictions

6. 📊 Model Performance Summary

Model	Accuracy	Precision	Recall	F1-score
Logistic Regression	0.769	0.750	0.815	0.781
Random Forest	0.795	0.805	0.784	0.794
XGBoost	0.800	0.794	0.814	0.804
Neural Network	0.802	0.791	0.827	0.808

🏆 Best Model: The Neural Network achieved the highest accuracy of 80.2%, along with the strongest F1-score (0.808), making it the most effective model for this classification task in our pipeline.

6. 📤 Submission Generation

• Final prediction pipeline applied to the test set
• Submission file generated in proper Kaggle format

---

🧑‍💻 Collaboration & Tools

• Version Control: GitHub (branching, pull requests, versioning)
• Shared Environment: Google Colab for code collaboration
• Libraries: Pandas, NumPy, Matplotlib, Seaborn, Scikit-learn, XGBoost, TensorFlow/Keras, SHAP

---