Business Analyst (Student Project): Titanic Survival Analysis

Used classification and clustering to analyze survival determinants

• Situation: Analyzed Titanic passenger data to identify survival factors and support risk assessment strategies for modern shipping.

• Task: Determine key attributes influencing survival outcomes and model risk based on class, gender, and age.

• Action: Cleaned and prepared nominal dataset; applied J48 decision tree (C4.5) and KMeans clustering in WEKA to uncover patterns; tuned decision tree parameters for more detailed granularity.

• Result: Achieved 79% classification accuracy, ROC AUC = 0.82; revealed that first-class status and gender (female) significantly increased survival odds. Clustering validated class and gender as critical survival factors, supporting more equitable evacuation planning.

  
  

Roc and confusion matrix of decision tree algorithm

Visualization of decision tree

Outcome of decision tree algorithm

WEKA Clusterer Visual.

The outcome of the clustering algorithm.