[Prev][Next][Index]

Projection for 17th observation



Dear Marco,

This is the second response to your mail dated July 1st. Based on your mail, I 
go back to the model and replaced C(15)=1120 with 895, thus the projection for 
K=16 would be:

P(16)
A=485
B=245
C=812
D=1494
E=369

The complete output table follows.
Table of MDSM output for 15th observation and 16th projection.
k=15					
    P(15)      D(15)    E(15)     IV(15)     R(15)     T(15)      P(16)
						
A   597.5941   520     535.5188   0.292641  -62.0753  0.931759    484.5149
B   255.0158   234     238.2032   0.130169  -16.8127  1.046004    244.7648
C  1133.465    895     942.693    0.515146  -190.772  0.907379    812.1038
D  1408.263   1404    1404.853    0.767698  - 3.4104  1.063965   1493.807
E   410.2986   370     378.0597   0.206595  -32.2389  0.997123    368.9356
   1964.292  1798.415 1829.954    1         203.9142  2.216283   1822.567
                0.9955    0.9967				
                          1.0012				

Furthermore, I entered 16th observation, as D(16), based on your email, and the 
MDSM projected P(17) for next observation which due the end of July:

P(17)
A=522
B=236
C=929
D=1448
E=330

The complete output Table follows.
Table of MDSM output for 16th observation and 17th projection.
k=16					
     P(16)     D(16)    E(16)    IV(16)     R(16)       T(16)    P(17)
						
A    484.5149   535    524.903  0.285692   40.38812    0.976256   522.2971
B    244.7648   237    238.553  0.129839  - 6.21186    0.997463   236.3988
C    812.1038   952    924.0208 0.502923  111.9169     0.976272   929.4113
D   1493.807   1422   1436.361  0.781777  -57.4459     1.018339  1448.079
E    368.9356   352    355.3871 0.193429  -13.5485     0.936269   329.5668
SGI 1822.567   1842.4671837.302 1         132.9616     2.19425   1843.375
	0.9997	0.9974				
		0.9977				

Let's see how close the projection will be this time.

P.S. You mentioned in your last mail that you were also collecting weekly-data. 
Some times changing the time interval makes a better fit. Considering your data 
are microorganisms, weekly data may describe the dynamics better.

Although a better projection would generally catch people' attention, I would 
not expect this model to make a perfect projection. The purpose of this model 
is only for supplying a formula or quantified mechanism that allow us to 
monitor dynamics of communities. It is expected that this model would improve 
and extend our understanding of the dynamics of multivariate communities.

P.P.S. After carefully read your mails, I understand that the 15th observation 
was collected June 3rd, but posted June 19th. Ignore my question in previous 
mail.

Thank you,

T. Jay BAI, Ph.D.
Quantitative Ecologist
970-490-8345
Bai@gpsr.colostate.edu
http://lamar.colostate.edu/~jbai