π LARGE-SCALE DATA STORYTELLING
π LARGE-SCALE DATA STORYTELLING
Embedded Files
π Completed in under 2 hours as part of a timed visual analysis challenge.
Finding the Reliable Data: Revealing the True Story from 86,635 Data PointsΒ
Finding the Reliable Data: Revealing the True Story from 86,635 Data PointsΒ
What I did: Revealed that surface metrics, often assumed to be the most reliable, turned out to be the most volatile β a risky foundation for decision-making.
Why it matters: Shows how fast, targeted EDA can:
β’ Challenge default assumptions
β’ Stress-test the stability of KPIs
β’ Pinpoint where meaningful patterns actually live
What I used: R, tidyverse, ggplot2, stratified time-depth analysis, custom visual storytelling
VizPosterF1_Portfolio_Kelly.pdfVisual Forecasting from 700,000+ Records of F1 Performance DataΒ
Visual Forecasting from 700,000+ Records of F1 Performance DataΒ
What I did: Cleaned and standardized 11 seasons of raw racing data (2014β2024) to engineer KPIs like podium share, win rate, and team consistency. Built ranked visuals to identify top-performing teams for the upcoming season.
Why it matters:
Shows how to extract decision-ready insights from chaotic, multi-source data
Demonstrates KPI design, trend analysis, and stakeholder-driven storytelling
Highlights my ability to turn raw data into strategic visuals
What I used: R, tidyverse, ggplot2, multi-table joins, KPI engineering, trend analysis
β¬οΈ Worked with 211,000+ global records to build high-level metrics and test whether action translated to real impact β surfacing gaps that raw data alone didnβt reveal. Β π‘GitHub
β¬οΈ Worked with 211,000+ global records to build high-level metrics and test whether action translated to real impact β surfacing gaps that raw data alone didnβt reveal. Β π‘GitHub
Focused on trends that show where the next big impact will come from.Β
Mapped growth against outcome to stress-test assumptions about impact.Β
Page updated
Google Sites
Report abuse