Most Frequent Crime Type vs. Moon Phase
Association Rules consist of the most frequent crime, different moon phase, sun hour level and average temperature level are generated individually with a minimum support of [0.03, 0.05, 0.08, 0.1, 0.3]. The minimum confidence is set at 0.3 and minimum lift at 1. If the lift is greater and 1, it may indicate that the items involved may have an underlying relationship.
The pattern that is the most frequent for this case is:
-
{SunHour_3, AvgTemp_4}, with support = 0.433532
This rule indicates that 43.35% of the time, when the Sun hour is in the middle level, it will associate with a relatively higher average temperature.
Other association rules with minimum support of 0.2 are as the following:
-
{Theft} -> {SunHour_3}, support = 0.285148, confidence = 1.0, lift = 1.0
-
{Waning Crescent} -> {SunHour_3}, support = 0.230344, confidence = 1.0, lift = 1.0
-
{Burglary} -> {SunHour_3}, support = 0.210244, confidence = 1.0, lift = 1.0
These rules indicate that more than 20% of the times, when the most frequent crime types are Theft and Burglary, the sun hour level of that day is more likely to be in the middle; and when the moon phase is Waning Crescent, the sun hour will also likely to be in the middle level.
Different Crime Types vs. Moon Phase
Association Rules with minimum support of [0.05, 0.08, 0.1, 0.2, 0.5] are generated individually with minimum confidence 0.3 and minimum lift of 1. Rules associate different crime types with different moon phase.
The pattern that is the most frequent for this case is:
-
{Theft_1, Assault_1}, with support = 0.987583
This rule indicates that 98.76% of the time, when the Theft crime occurrence is low, the occurrence of Assault crime will also be low. This rule intuitively makes sense since when the occurrence of one crime is low, it may indicate that this place/location overall is safer, and thus the likelihood of the occurrence of another crime will also be low.
Other association rules with minimum support of 0.9 are as the following:
-
{Robbery_1} -> {Theft_1}, support = 0.965197, confidence = 0.993418, lift = 1.004894
-
{Drug_1} -> {Fraud_1}, support = 0.953677, confidence = 0.984042, lift = 1.001289
-
{Fraud_1, Assault_1} -> {Drug_1}， support = 0.951419， confidence = 0.984097, lift = 1.001228
The above rules also imply that when an area has a overall low level crime occurrence of one type, it will also have low levels occurrence of other crime type. And certain crime types seem to associate with each other for more than 95% of the time, such as Assault and violence crime.
Association rules relate moon phase and crime types are as the following:
-
{Waning Crescent} -> {Fraud_1}, support = 0.226453, confidence = 0.983107, lift = 1.000338
-
{Waning Gibbous} -> {Robbery_1}, support = 0.128202, confidence = 0.972211, lift = 1.000638
This may infer that there may exist certain relationship between moonphase and certain types of crime.
Most Frequent Crime Type vs. School Type
This association rule considers the relationship between the most frequent crime type and different school types. With minimum support of [0.05, 0.08, 0.1, 0.2, 0.3], association rules are generated individually with minimum confidence 0.3 and minimum lift of 1.
The pattern that is the most frequent for this case is:
-
{Total_Pub_1} -> { Total_Pri_1 }, with the support = 0.713591
This rule indicates that 71.34% of the time, lower number of public schools in one place/location may associate with lower number of public schools in that area. So, for places with lower education level, low level of one type of school should associate with a lower number of another type of school.
Other association rules with minimum support of 0.2 are as the following:
-
{Theft} -> {Total_Pub_1}, support = 0.211224, confidence = 0.740751, lift = 1.010172
-
{Burglary} -> {Total_Pri_1}, support = 0.20072, confidence = 0.9547, lift = 1.004445
These rules indicate that when the most frequent crime types are Theft and Burglary, more than 20% of the times, that location would have lower numbers of public and private schools respectively.