Explainable AI (SHAP)
CreditSense AI
Credit Scoring Model
PythonXGBoostPandas
✦ WELCOME TO MY
Machine Learning
Laboratory
Explore interactive ML projects built with real-world datasets and modern algorithms. Test, learn, and experience the power of machine learning — live.
6+ML Projects
10+Trained Models
6Real Datasets
2.8K+Predictions
100%Live & Interactive
“
Real projects. Real models. Real impact. Interact, explore, and see machine learning in action..
— Zishan AhmadMACHINE LEARNING FOR A BETTER TOMORROW
Featured Projects
Click “Live Demo” on any project to test it immediately in the interactive playground below.
Binary Classification
Titanic Survival
Prediction
PythonRandom ForestSeaborn
Anomaly Detection
Credit Card Fraud
Detection
Pythonimbalanced-learnScikit-learn
Multiclass Classification
Iris Flower
Classification
PythonSVMMatplotlib
Linear Regression
Sales Prediction
(Regression)
PythonLinear RegressionPandas
Regression Scoring
Movie Rating
Prediction
PythonMLPScikit-learn
Model Online
Titanic Survival
Estimates a passenger's survival probability from class, age, fare, family size and title, deriving the engineered features server-side exactly as the notebook did.
Kaggle Titanic Dataset
Model Parameters
Prediction Result
The prediction result isSURVIVED
Probability of Survival99.71%
Model Confidence: Very High
Input Summary
Pclass:1st Class
Sex:Female
Age:28
SibSp:1
Parch:0
Fare:71.83
Embarked:S
What's Next?
Try different values, explore the model performance, or check the feature importance.
Why this prediction?Feature Importance
Sex0.42
Pclass0.24
Fare0.16
Age0.09
SibSp0.05
Parch0.02
Embarked0.02
Model Performance (Random Forest (100 trees))
84.03%Accuracy
83.10%Precision
81.23%Recall
82.15%F1 Score
0.90ROC-AUC
About This Project
The Titanic Survival Prediction project uses machine learning techniques to predict whether a passenger survived the Titanic disaster based on personal information. We use various classification algorithms and ensemble methods to build a robust predictive model.
Tech Stack
PythonPandasNumPyScikit-learnMatplotlibSeaborn
Curated Real-World Datasets
Every project in this lab is backed by verified public datasets with known origins, reproducible splits, and citation links.
CreditSense AI
Explore Dataset ↗German Credit & Lending Club Data
50,000+ verified records
Titanic Survival
Explore Dataset ↗Kaggle Titanic Dataset
891 verified records
Credit Card Fraud
Explore Dataset ↗Kaggle Credit Card Fraud Dataset
284807 verified records
Iris Flower
Explore Dataset ↗Fisher's Iris Dataset
150 verified records
Sales Prediction
Explore Dataset ↗Advertising Budget & Sales
200 verified records
Movie Rating
Explore Dataset ↗IMDb Indian Movies Dataset
15509 verified records
The End-to-End ML Pipeline
How models transition from raw Kaggle data through engineering and training into live production inference.
- 01
Data collection
Public datasets — Kaggle sources, cited on every model.
- 02
Preprocessing
Cleaning, encoding and scaling, fit on the training split only.
- 03
Training
Model fit with cross-validation; hyperparameters tuned.
- 04
Evaluation
Held-out test metrics — the same numbers shown on each model.
- 05
Deployment
Exported artifact served behind the API you run here.
Machine Learning Engineering Stack
Modern Python, Scikit-learn, XGBoost, and FastAPI backend infrastructure combined with Next.js and React.
Python 3.13Language
Scikit-learnCore ML
XGBoostGradient Boost
SHAPExplainability
Pandas & NumPyData Science
FastAPIInference API
DockerContainerization
Next.js 15BFF & SSR
Tailwind CSSStyling
LenisSmooth Scroll
RenderCloud Hosting
GitHub ActionsCI / CD
About the ML Lab
Created by Zishan Ahmad, AI & ML Engineer based in Prayagraj, India. This laboratory serves as an open, interactive showcase of production machine learning systems. Each model is trained end-to-end, validated against held-out test splits, and engineered for real-time inference.