Steven R
Wilson

Data Analyst | Statistical modeling for capacity and throughput questions

Projects

Real questions,
data worked end to end

Illustrated aerial view of a college-town neighborhood at sunset, overlaid with a scatter trend line, a histogram and heatmap tiles

Ames Housing MLR

Does the same regression, built in R and in Python, land on the same answer?

$53 RMSE gap, R vs Python
7.7% Test-set MAPE
View project →
Illustrated freeway at dusk, with free-flowing lanes beside congested ones under a speed heatmap ribbon

Traffic Congestion Analysis

What separates a slow freeway from one that is actually breaking down?

5-min Loop detector resolution
3 definitions 1,457, 10 and 7 events
View project →
Illustrated bike shop with a road bike, shipping boxes and apparel, overlaid with line, bar and donut charts

SQL Business Metrics

Which sales channel actually drives revenue, and why does order volume mislead?

87.9% Online share of orders
26.7% Online share of revenue
View project →
Illustrated electricity demand curves over a California landscape at dusk

Time Series Forecasting

When during the day does the official demand forecast miss, and can simple models do better?

84% Of EIA shortfall, 8am–4pm
11% Gain over seasonal naïve
View project →
Illustrated customer clusters as four colored point clouds with centroids, beside faint histograms

Cluster Analysis

Which customers actually carry the business, and what does RFM scoring miss?

72.8% Revenue from 18% of accounts
5,823 Accounts clustered
View project →
Illustrated In-N-Out restaurant with a drive-thru line and busy counter, overlaid with line and bar charts

In-N-Out Business Data

How fast does a drive-thru lane actually clear, and where does the line stall?

HH:MM:SS Hand-logged in store
2 modes Drive-thru & counter
View project →
[ALT TEXT: describe the NASA hero image]

NASA SOFIA Internship

In summer 2013 I joined NASA’s SOFIA program with no assigned project, so I found one: an engineering problem aboard the airborne observatory. I built the analysis in R, flew on two observing missions, and turned the work into a research poster.

2 Observing missions flown
R Analysis built in-house
View project →

About

How I work

My projects start with a question, not a method.

I compare models, test their assumptions, look for where those assumptions break down, and say plainly what the results do — and do not — support. I’m most interested in what happens when a model meets messy real-world data: imputed detector readings, anomalous source values, data I collected myself.

I have a BA in Mathematics from Cal State Fullerton and spent a summer at NASA using R for time-series analysis of instrument data across multiple missions. I work with R, SQL, and Python.

Contact

Get in touch