After confirming stationarity, I was ready to construct a model to forecast police shootings. I chose the SARIMAX model for its ability to handle both seasonal and non-seasonal elements. Here’s a snapshot of what I found:
Model: SARIMAX(1, 1, 1)x(1, 1, 1, 12)
AIC: 12537.839
BIC: 12568.175
The AIC and BIC values were my benchmarks for model performance – lower numbers are generally better, and they suggested that my model was a snug fit: complex enough to capture the trends and patterns without overfitting the data.
But numbers aside, the real story was in the coefficients, particularly for the moving average terms. These coefficients were significantly negative, hinting at a strong influence from recent events on the likelihood of future incidents.
What I’ve developed is more than just a statistical model; it’s a potential crystal ball into the when and where of police shootings. This isn’t just academic – it’s about real lives and the hope that, by predicting these events, we might find a way to prevent them.
Be First to Comment