saketh guptha

Downloaded six NHANES datasets from the CDC website and used Python (Pandas) to merge survey responses, accelerometer readings, demographics, and health outcomes into a single, schema- consistent table of 5,326 participants (61,168 records), tracking data provenance throughout.

Developed a modular Python cleaning pipeline to impute missing values, convert float fields to the correct numeric types, bucket ages into uniform groups, and standardize column data types—version-controlled for reproducibility and peer review.

Conducted exploratory and inferential analyses (descriptive statistics, grouped summaries, t-tests, ANOVA) and produced publication-quality visualizations (heatmaps, line/bar/cluster plots) to inform iterative model design and stakeholder reviews.

Transformed a five-level self-reported health scale into a statistically validated binary “good” vs. “poor” label, strengthening the target signal and ensuring model robustness.