HOVA-FPPM: Flexible Periodic Pattern Mining in Time Series Databases Using Hashed Occurrence Vectors and Apriori Approach

Algorithm 1

The proposed algorithm.

	Input: event E, Occ_vec All Events S, Max_Pattern_Length, Prev_Keys
	Output: a list of mined patterns
	NP = {}, New_Events = slice_even (S, Prev_Keys) Periodic = periodicity (E, Occ_vec,lengthofOcc_vec,Confidence) for i in {O1, ..., On} of Period do
	for Key in {K1, ..., Kn} of New_Events do for Item
	in {I1, ..., In} of
	New_Events(Keys) do
	Difference ← Periodic − Item if
	continue
	end   Difference > Max_Pattern_Length
	then
	if Item > Periodic then if
	Difference > KeySize then
	star_count = Difference−
	KeySize
	end
	if star_count > star_limit then
	end
	continue
	end
	end   stars = calculate_stars(star_count) pattern_key = E + stars + Key
	end   patterns_with_stars = Periodic[i] NP ∪ pattern_with_star
	else
	pattern_key = Key + E patterns_without_stars = Periodic[i]
	NP ∪ pattern_without_star_count
	end  if Difference > KeySize then
	star_count = Difference − KeySize
	else
	if star_count > star_limit then
	continue
	end   end
	end    stars = calculate_stars (star_count) pattern_key = Key + stars + E pattern_with_star
	end    =Periodic[i] NP ∪ pattern_with_star
	pattern_key = Key + Epattern_without_star = Periodic[i] NP ∪
	pattern_without_star
	end for j in {Key1, ..., Keyn} of New_Events do
	NL_Patterns = calc_next (New_Events, NP, Max_Pattern_Length) NP ∪
	update (NL_Patterns)
	end
	end
	ReturnNP