These rules are generated using the apriori
function of the arules
package. Let's generate the rules for the other dataset this time:
rules2<- apriori(sampdata,parameter = list(sup = 0.45, conf = 0.9, target="rules"));
The following is the output of the preceding command:
We set the threshold for support
as 0.45
and the threshold for confidence
as 0.9
. From the preceding output, we can see that there are about 40
rules generated. We can print them in descending order of their lift
ratio using the following code:
inspect(head(sort(rules2, by="lift"),10))
The output is as follows:
From the output, it is clear that whenever the i128
and i141
items co-occur, it is most likely that the i139
item will occur. Additionally, as the lift
value is more than two, it further reiterates that the combination is most likely to occur. We can also get the top rules based on the combination of support
, confidence
, and lift
using the quality
function:
head(quality(rules2));
The output...