Human Body Temperature Analysis - Statistical Inference
MY ROLES
TIMELINE
2019
TOOLS USED
Statistical Inference, Hypothesis Testing, Confidence Intervals, Python, Pandas, Matplotlib, Bootstrap Analysis, t-tests, Normal Distribution, Medical Research
DESCRIPTION
This project conducted a rigorous statistical analysis to examine whether the widely accepted 'normal' human body temperature of 98.6°F (established in 1868) is statistically accurate. Using both bootstrap and frequentist approaches, the analysis tested the distribution of body temperatures, evaluated the true population mean, identified thresholds for abnormal temperatures, and investigated gender differences in normal temperature.
OUTCOMES
- Demonstrated that normal body temperature is not 98.6°F but closer to 98.2°F using statistical testing
- Identified gender differences in normal body temperature with statistical significance
- Established evidence-based thresholds for abnormal temperature using confidence intervals
FEATURES
Distribution normality testing, bootstrap hypothesis testing, t-tests and z-tests comparison, confidence interval calculation, gender-based analysis
CHALLENGES
Working with a controversial scientific standard that had been accepted for over 120 years. Determining which statistical tests were most appropriate for different sample sizes. Explaining complex statistical concepts in accessible medical contexts.
APPROACH
Conducted normality tests on body temperature data to verify distribution assumptions. Applied both bootstrap and frequentist approaches to test the 98.6°F benchmark. Compared results from different sample sizes to illustrate statistical principles. Calculated confidence intervals to determine abnormal temperature thresholds. Used appropriate testing methods to identify significant gender differences in normal temperature.