Using the data of Fig. 4, the 12 types of traffic violations, "speeding", "not wearing seatbelt", "ignoring a stop sign", "illegal parking", "drunk driving", "passing where prohibited", "ignoring a traffic light", "driving in the wrong lane", "crossing where prohibited", "neglect of the railroad crossing rule", "cell-phone use while driving", and
"poor maintenance" were compressed into four components to study violation patterns with PCA(principal components analysis).
Fig. 5 displays the relationship of the original 12 offences and the four components of compressed data.
Red arrows indicate a positive correlation, while blue arrows show a negative correlation.
The first component represents violations that are committed most often, and is entitled "repetitive offences".
The second component consists of violations that tend to be intentional, such as "ignoring a no-crossing sign" and "illegal parking", and is therefore dubbed "willful offences". The third, more often committed unwittingly rather than intentionally, is referred to as "inadvertent offences". The fourth is strongly related to the use of alcohol, and is called "alcohol-related offences".
In this analysis (PCA), violation data that are not fully explained in the first component are compressed into the remaining three components. While the first component illustrates quantitative abundance, the rest of the components contain qualitative information concerning violations not fully explained by the first compressed component.
|
