Probability model
Coefficient estimates of the developed logistic regression
model for the growth of A. flavus are presented in Table 6. The
model included %mc, temperature and time observations; taking
into account the full matrix the model predicted correctly 91% of
the cases, with 4.7% false positives and 4.3% false negatives (cut
off ¼ 0.5). Most of wrong predictions were due to a deviation in
the predicted delay periods before growth, but growth/nogrowth
conditions (%mc/temperature) in the long term were
correctly predicted (cut off ¼ 0.5, 30 days) in all cases except for
20 C/15%mc where the observed probability was 0.60 and the
predicted one 0.34. Probability of growth for different %mc and
temperature values at 30 days of incubation and along 60 days for
a mc ¼ 15% are presented in Fig. 4. It can be graphically depicted
that probability increased with time and with %mc level. Predicted
probabilities at time 0 could be as high as 0.99 in those
conditions where growth was detected from day 1, while it
decreased with less growth-conducive conditions; a high probability
at time 0 must not be interpreted as positive growth at the
inoculation time but at the first-second day of incubation. At
temperature values near the optimum for growth (30 C) A. flavus
probabilities for growth over 0.50 were predicted at 13%mc after
30 days. As observed in the figures, probabilities of growth for
A. flavus over 0.80 were predicted over 14%mc at 25e33 C after
30 days. Thus, for safe pistachio storage, long time periods of
exposition to temperatures between 25 and 33 C should be
avoided. On the other hand, probabilities
Feel lonely, no happiness.Feel lonely, no happiness.Feel lonely, no happiness.Feel lonely, no happiness.Feel lonely, no happiness.Feel lonely, no happiness.Feel lonely, no happiness.Feel lonely, no happiness.
การแปล กรุณารอสักครู่..
