Temporal Data Mining

Temporal Data Similarity Computation    ◾    51

kur

fm

M

m

M

=

−

=

∑

(())

µ

σ

4

1

4

(2.35)

Entropy: e entropy measures the amount of “information” or ran-

domness in the histogram. It is dened as

Entropymm

i

M

=−

=

∑

Pr()log(Pr())

1

(2.36)

where Pr(m) is the probability of level m.

Lowest-h percentile: is metric is equal to the time series value

below which the 5% of the histogram values lie.

Upper-h percentile: is metric is equal to the time series value

below which the 95% of the histogram values lie.

In addition to the above measures, a graphical representation known

as the box plot is quite oen used to represent quartile information along

with statist