Abstract
Eighty-one prescription drugs were divided into three groups according to the strength of the auto-correlation (strong, weak or none) of daily variations in the drug supply amounts at a pharmacy. The power spectral density and auto-correlation function were used as an indicator for the classification. Sixty-four drugs fell into the no auto-correlation group, 15 drugs into the weak auto-correlation group and 2 drugs into the strong auto-correlation group. Interestingly, our grouping using auto-correlation as a criterion was consistent with the target-oriented classification (standard commodity classification of Japan).
