I’m back with an update on Pedestrian-Traffic Incidents for the month of January. The data is sourced from the City of Calgary’s open data website. The data set is here.

There were 32 incidents in January – 8 more than in December, and 19 more than this time last year.

The heat map and cluster map are largely unchanged, even adding 32 incidents to the existing 1288 is not enough to make any noticeable changes.

There are still distinct clusters downtown (red), Forest Lawn (orange), along the Northeast transit line (green), and Macleod Trail (blue).

Model Details

The model is a SARIMAX (Seasonal Auto-Regressive Integrated Moving Average eXogenous regressors). The model specification is (0, 1, 1)(0, 1, 1)[12]. An Error-Trend-Seasonal breakdown is shown below:

Regression diagnostics are shown below:

The regression diagnostics indicate that the model is not missing any major dynamics in the data. The standardized residuals are centered on zero, and do not show significant autocorrelation. As with last month, the residuals are more bimodal than in the past, this shows up in the histogram, and as a deviation from the linear pattern in the Q-Q plot at the -1 and +1 Theoretical Quantiles. None of the lags on the Correlogram are statistically significant.