Merge branch 'main' of https://github.com/NewYorkCityCouncil/jaywalki…

…ng_2024

README.md

@@ -31,7 +31,7 @@ The data team analyzed citation and crash data from NYC and other cities in orde
 - **We found weak to moderate statistically significant positive relationships between the quarterly number of criminal summonses issued for jaywalking and pedestrian-involved/ jaywalking-involved collisions (with a one quarter lag), suggesting that summonses and incidents rise and fall together.** A negative relationship, where increases in summonses issued correlate with decreases in accidents and vice versa, would have been much more suggestive of jaywalking summonses protecting against accidents. 
 - **In NYC, there is a statistically significant moderate relationship between pedestrians killed and injured (PKI) and jaywalking summons issued per 10,000 people per precinct. This suggests that there is some relationship between the sites of severe accidents and the locations targeted with jaywalking summonses.** However, some precincts, like 9 (MN), 28 (MN), 33 (MN), and 76 (BK), are exceptions, as they fall in the top 25th percentile for their share of the city’s jaywalking summonses and the bottom 25th percentile for their share of the city’s total PKI.
 - 52% of all pedestrian-related car crashes occur in Vision Zero Priority Zones, and 58% of jaywalking summonses are given in priority zones. **However, there is a strong racial disparity in these regions. The priority zones are about 73% non-white, but 92% of jaywalking summonses in these areas are given to non-white people.**
-- **The results of our Granger causality test revealed that the number of citations in the previous quarter is associated with changes in the pedestrian-involved share of collisions in the next quarter. However, it is important to note that Granger causality does not necessarily imply true causality.** Granger causality is rooted in predictability and correlation rather than a direct cause-and-effect relationship. **In the case of citations predicting the share of pedestrian collisions, it is especially unlikely to indicate true causality** given the effect is small (β = -0.0039) and the model is not very predictive. The mean magnitude of error is 124% indicating that the error is often larger than that actual value itself.  
+- **The results of our Granger causality test revealed that the number of citations in the previous quarter is associated with changes in the pedestrian-involved share of collisions in the next quarter. However, it is important to note that Granger causality does not necessarily imply true causality.** Granger causality is rooted in predictability and correlation rather than a direct cause-and-effect relationship. **In the case of citations predicting the share of pedestrian collisions, it is especially unlikely to indicate true causality** given the effect is small (β = -0.0039) and the model is not very predictive. The mean absolute percentage error of the model is 124% indicating that the error is often larger than that actual value itself.