shouldibuygastoday

Model

WTI Gas Timing Model

Predicts whether consumers should buy gas now based on recent WTI crude oil price levels and short-term trends.

Accuracy48%
AlgorithmLogisticRegression · C=1 · l2
DataEIA — Petroleum Spot Prices (WTI)

Decision surface

The model's buy probability across its two most influential features. Every point was computed by calling model.predict_proba on the actual .joblib. Drag to rotate.

What the model pays attention to

wti_price_7d_avg41%
wti_price_30d_avg30%
wti_price19%
wti_price_change_7d10%

Approach

I used daily WTI crude oil spot prices from the EIA API to train a logistic regression classifier. I engineered features that capture current crude price, short-term average price, recent 7-day change, and a 30-day trend. The target is whether the average crude oil price over the following week is higher than the current price, which serves as a practical proxy for upcoming upward pressure on retail gas prices. I chose logistic regression because it is simple, interpretable, and works well for binary classification while supporting both predict and predict_proba for the team pipeline.

Features

wti_pricefloat
Current WTI crude oil spot price ($/barrel)
wti_price_7d_avgderived
7-day rolling average of WTI price
wti_price_change_7dderived
Change in WTI price over the last 7 days
wti_price_30d_avgderived
30-day rolling average of WTI price

Built by

Christa IrakozeLinkedIn

Business Analytics graduate student at Arizona State University with a background in data analysis and reporting automation. I am interested in building practical machine learning solutions that connect economic signals to real consumer decisions, especially in energy and pricing trends.

Trained on

1986-02-12 to 2005-10-18

All models