SVSatyavamsi GitHub ↗
Data Science · Analytics · Machine Learning

Turning data intoclear insights,models & products.

I work across analytics, data science and machine learning — exploring data, finding useful patterns, building predictive models and turning results into clear, usable products.

View projects Explore GitHub ↗
30Kcredit records analyzed
541Kretail transactions explored
2live data products
01 / ABOUT

I use data to answer questions, explain patterns and build useful predictive systems.

My work spans exploratory data analysis, business-focused analytics, statistical reasoning, visualization, forecasting and machine learning. I care about understanding the problem and communicating the result just as much as training the model.

The goal is simple: turn raw data into work that another person can understand, reproduce and actually use.

02 / PROJECTS

Featured work

A growing collection across analytics, forecasting, data science and machine learning.

Credit Default Risk Intelligence● LIVE
Risk Assessment

Default Risk Prediction

Interactive portfolio demonstration using a trained gradient-boosting model.

26.5%threshold
ROC-AUC0.780
PR-AUC0.553
Recall0.582
Open live app ↗
PROJECT 01 LIVE

Credit Default Risk Intelligence

End-to-end credit-risk classification system using 30,000 historical client records. Includes leakage-safe preprocessing, model benchmarking, five-fold cross-validation, threshold tuning, permutation importance, subgroup auditing and a deployed Streamlit interface.

ROC-AUC0.780
PR-AUC0.553
Recall0.582
F10.546
Pythonscikit-learnXGBoostStreamlitGitHub Actions
Live demo ↗ Source code ↗

Educational portfolio project based on historical UCI data; not a production lending decision system.

Retail Sales Forecasting Intelligence● LIVE
Analytics + Time Series

Retail Revenue Forecasting

Transaction analytics and chronological forecasting across more than half a million retail records.

Transactions541K
MAE£14.9K
Horizon42 days
Open live app ↗
PROJECT 02 LIVE

Retail Sales Forecasting Intelligence

Business analytics and time-series forecasting using the UCI Online Retail dataset. Cleans 541,909 transactions, analyzes revenue patterns, engineers lag and rolling features, and evaluates forecasts on a 42-day chronological holdout.

Transactions541K
MAE£14.9K
RMSE£27.9K
MAPE26.3%
PythonTime Seriespandasscikit-learnStreamlitAnalytics
Live demo ↗ Source code ↗

The 7-day seasonal baseline beat the more complex ML models on MAE and RMSE, showing why strong forecasting baselines matter.

03 / SKILLS
Analytics

Insights & visualization

Exploratory analysis, KPI thinking, data cleaning, visualization and clear communication of findings.

Data Science

Statistics & forecasting

Feature engineering, statistical reasoning, time-series analysis, forecasting and experiment design.

Machine Learning

Modeling & deployment

Classification, regression, cross-validation, tuning, explainability, Streamlit and reproducible workflows.

PythonSQLpandasNumPyMatplotlibscikit-learnXGBoostTime SeriesStreamlitGit
04 / CONTACT

Have a project, idea or opportunity?

Explore my work and get in touch through GitHub.

Visit GitHub ↗