• tomaz.kastrun - Saturday, January 21, 2017 9:02 AM

    Jonathan Mallia - Saturday, January 21, 2017 6:49 AM

    Amazing article!

    Can you please shed some light on confidence and support for the last part of the article, how they are related directly or indirectly, and what difference does it make when you adjust their values in arules ?

    Thanks a lot!
    Jon

    Hi,
    Both support and confidence are important to identify and find relevant relationship between left hand side (LHS) and right hand side (RHS). Left hand side is interpreted as IF item A.... and Right hand side as THEN item B and item C.
    Or shown graphically {A}  => {B,C}. Imagine, this is our rule. To this rule, support represents, how many times this rules was found in the dataset. If support is 0.123, this means that this rules appeared 12,3% out of all the rules in dataset.
    Confidence will tell you, how many times this rules has been proven as True. If confidence for our rules is 0.99, this means that in 99% of the dataset containing all the rules, customers that bought item A will in 99% times also buy item B and C.

    Best, Tomaž

    Thank you Tomaz,

    it's a lot clearer with your explanation.

    Many thanks.