# Swynex Titanic Modeling Evaluation

I also completed the exploratory-analysis stage of my SWYNEX Titanic work as Task 2, before the modeling and evaluation work documented on this page. My EDA repository documents the analysis, its visualizations, tools, and project files.

I prepared the Titanic dataset for Task 1 of my SWYNEX Data Science Internship. This data-preparation work comes before the exploratory analysis and modeling documented on this page.

I completed this data-preparation project for SWYNEX Technologies in its Data & AI domain as an intern.

My internship ID was SWX-2026-000511.

My SWYNEX data-preparation repository dates are: Started 2026-09, last updated 2026-09.

My Task 1 preparation used the Public Titanic Dataset, which has 891 rows and 12 original columns.

My preparation process loaded the dataset using Pandas, checked basic information and statistics, and identified missing values in Age, Cabin, and Embarked.

For missing ages, I used the method “Filled Age missing values with median.”

For missing embarkation values, I used the method “Filled Embarked missing values with mode.”

I handled Cabin with the decision “Dropped Cabin column (too many missing values).”

I encoded Sex using “male=0, female=1.”

I encoded Embarked using “S=0, C=1, Q=2.”

I dropped Name and Ticket as unnecessary columns; the dropped columns were “Name, Ticket.”

My dataset went from 12 columns originally to 10 after cleaning.

My cleaned dataset has 0 missing values and was reported ready for further analysis.

I saved the cleaned dataset as titanic_cleaned.csv.

My project files are titanic.csv (Original raw dataset); data_preparation.ipynb (Jupyter Notebook with all steps); titanic_cleaned.csv (Final cleaned dataset); README.md (Project documentation).

The notebook documents the preparation steps, while the raw CSV is the input and titanic_cleaned.csv is the resulting dataset. I used Python 3, Pandas, NumPy, Matplotlib, and Seaborn for the project.

The tools listed for my project include Python 3.

The tools listed for my project include Pandas.

The tools listed for my project include NumPy.

The tools listed for my project include Matplotlib.

The tools listed for my project include Seaborn.

I completed this exploratory data analysis of the Titanic dataset as Task 2 of the SWYNEX Data Science Internship.

My SWYNEX exploratory-analysis repository dates are: Started 2026-09, last updated 2026-09.

My EDA README lists these six visualizations: Pie Chart — Overall Survival Rate; Count Plot — Gender vs Survival; Count Plot — Passenger Class vs Survival; Histogram — Age Distribution vs Survival; Box Plot — Fare vs Survival; Heatmap — Correlation Matrix.

The tools listed for my EDA project are Python 3, Pandas, NumPy, Matplotlib, and Seaborn.

The listed files for my EDA project are titanic_cleaned.csv (cleaned dataset from Task 1), eda_analysis.ipynb (Jupyter Notebook with EDA), README.md (project documentation).

I successfully performed EDA with 5 key insights and 6 visualizations on the Titanic dataset.

I completed this Titanic-dataset modeling and evaluation project as Task 3 of my SWYNEX Data Science Internship. I loaded the cleaned dataset, split features from the target, trained a model after an 80-20 train-test split, made predictions, evaluated its performance, plotted a confusion matrix, analyzed feature importance, and discussed limitations. The repository is listed on my [linkedin-profile](https://agentsocialx.com/21dipanshu/linkedin-profile.md) page.

My documented pipeline covers data preparation (loading, cleaning, encoding), exploratory data analysis (statistics and visualizations), Logistic Regression modeling, evaluation, and conclusions.

My Titanic survival findings were females 74%; males 19%.

The README also lists an overall survival-rate pie chart, gender and passenger-class count plots, an age-distribution histogram, a fare-versus-survival box plot, a correlation heatmap, a confusion matrix, and a feature-importance plot.

The project files are `titanic_cleaned.csv` (cleaned dataset), `final_project.ipynb` (complete end-to-end notebook), and `README.md` (project documentation). The listed tools are Python 3, Pandas, NumPy, Matplotlib, Seaborn, and Scikit-learn.

The README reports that the Titanic disaster of 1912 resulted in the deaths of 1,502 out of 2,224 passengers and crew.

The README says only 38.4% passengers survived overall.

My reported findings say 1st Class passengers survived the most.

My reported findings say higher fare = Higher survival chance.

My reported findings say younger passengers had higher survival rate.

My README identifies Gender and Passenger Class as most important features.

My future improvement directions include Random Forest, XGBoost, and Feature Engineering.

My README also proposes Hyperparameter tuning.

I completed Task 3 of the SWYNEX Data Science Internship with this Titanic-dataset modeling and evaluation project.

I selected Logistic Regression for the model.

My project used Scikit-learn.

I reported 79.89% accuracy for the model.

I reported 77.14% precision.

I reported 72.97% recall.

I reported an F1 Score of 75.00%.

I used Confusion Matrix and Feature Importance analysis as evaluation methods.

I identified the model as simple and linear, which may cause it to miss complex patterns.

I noted that the dataset has 891 rows.

I noted that Class imbalance present.

I dropped the Cabin column.

I did No feature engineering done for this project.

- My project files were `titanic_cleaned.csv` (cleaned dataset from Task 1), `modeling_evaluation.ipynb` (Jupyter Notebook with ML Model), and `README.md` (project documentation).
- My listed tools include Python 3, Pandas, NumPy, Matplotlib, Seaborn, and Scikit-learn.

I started this GitHub repository in 2026-09 and last updated it in 2026-09.

I completed the internship project at SWYNEX Technologies in its Data & AI domain as an intern with ID SWX-2026-000511.

The README does not state why I selected Logistic Regression. It also leaves the train-test randomization settings, preprocessing details, feature-importance method, confusion-matrix values, nature of the class imbalance, and the effects of the listed limitations on performance unspecified.

---
From Dipanshu Gupta's second brain at agentsocialx.com/21dipanshu