[Prev][Next][Index]

FYI: A Sample of Multivariate Temporal Dynamic Analysis



Dear Colleagues,

Attached is a sample of multivariate temporal dynamic analysis.
When I was preparing this mail, I have already received a response
from Mr. Marco Canepa from Italy. He said in his mail:
" ..the 15th observation data are:
A:    520
B:    234
C:  1120
D:  1404
E:    370
I immediately noticed a correct prediction for C and D,
A is correct as far as trend, B and E are not very correct.."

The multidimensional sphere model (MDSM) was originally developed
for vegetation data. Thanks Mr. Marco, who shed a light to the
model: it could also be used on microbiology or entomology for data
synthesis and analysis. Data in these fields are more accurate and
with a shorter time intervals. It might be easier to conduct a time
series analysis in these fields. Thus, I posted my communications
with Mr. Marco here to share with you what we discovered and am
looking for more comments and application opportunities.

Thank you for your attention and interest.

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

==========================
Dear Marco,

Please find enclosed multivariate temporal dynamic analysis output
that you requested.

Your variables A, B, C, D, and E were represented by numbers, 1, 2,
3, 4, and 5, respectively. I am not sure what they represent.  Just
as an example, I assumed that these species were plant species, and
the values were their productions, and I further assumed that your data
were vegetation data, i.e., all five species were collected from the
same community. Under these presumptions, I performed a multivariate
temporal dynamic analysis. If these data were other ecological data,
then you can convert them to responded species accordingly.

There are two tables attached. The first table, MarcoData, is your
original data. The second table, Marcoout, is the output of the
computer software of MultiDimensional Sphere Model (MDSM).
The MDSM treats your vegetation data as 5 component vectors, 5-vector,
and every species as a component of the vector. For example, the five
values, 504, 280, 1453, 1344, and 368 together made the 5-vector that
represents the vegetation of the first year. In other words, the
column D is a 5-vector representing the observation of the vegetation
at  given year, and same as the column P (projection), E (expectation),
R (prediction error), and T (multivariate instantaneous trend, see below).

There are 14 identical tables inside the Marcoout for each year of the
14 years. There are eight columns and they are explained briefly below:

The first column is the variable names, such as plant species, 1, 2, 3,
4, and 5; or V-SUM, Cos, and SMC.
P: projection values of the given year based on the previous information.
(But for the first year, the projection values are the same as the original
vegetation observation data.)

D: Observation Data. These are the values that the program read from
the input vegetation data that you supplied.

E: Expectation. Expected true values of the vegetation at given year
(E=P+D, it is weighted averages from projections and observations).

IV: Importance Values. It is the importance values, or relative composition,
of the plant species in the vegetation.

R: Projection Error. (R=E-P. Please notice that, the error definition is
the difference between the projections and expectations, instead of
between the projections and observations.)

T: Multivariate Instantaneous Trend.
[T=IV(k)/IV(k-1). This model analyzes the composition change of the
vegetation. It considers the composition change is the essential change
of the vegetation, or any community.]

P(k+1): Vegetation Projection for the next year.
(k is the index of the year.)

The first five rows of the output describe the five species. The last
three rows describe the general condition of the vegetation.

V-SUM: vector sum, or vector lengths, describe the general situation
of the vegetation

Cos: Cosine values between the two vectors describing the correlation
between the two vectors. (The Cosine values under column D express the
correlation between the observations: cosine<D(k), D(k-1)>, while the
values under column E express the correlation between the observation
and expectation: cosine<D,E>.)

SMC: System Monitoring Coefficient, SMC=Cos<E,D>/Cos<D,D>. When the
SMC value is greater than or close to one, we consider that the model is
working for system monitoring and it's projections were over fitting or
correct fitting the observations.

The lowest SMC value during the 14 years is the year of 11: 0.9981,
and the highest SMC values occurred on year of 10: 1.0030. As both of
them are not much less than one, we consider the projections are
fitting with the observations.

In your original email, you asked that:" What would the next sequence
of number will be ?" As a temporal dynamic analysis model, the MDSM
projected the values for the 15th year's vegetation. The values for
the plant species 1, 2, 3, 4, and 5, are:

597.59, 255.02, 1133.47, 1408.26, and 410.30, respectively.

How close is this projection? We have to wait till next year to
find out. But from the history of the analysis, i.e., the year of
2-13, we can expect that it would be a very close projection.

For your second question: "Is there a software available able to work
this out?"  The answer is yes. A DOS executable computer program,
SMM52 based on MDSM performing temporal dynamic analysis, can be
down loaded from my webpage:

http://lamar.colostate.edu/~jbai

(You can just click the webpage address, instead of typing it, as you
mentioned that there is no "wave" key on your keyboard. Furthermore,
there are some other references about this model that you may be
interested to check out from the web page. )

If you have any more questions, please contact us or send your message to

MDSM@grpsr.colostate.edu

Best Wishes,


T. Jay Bai, Ph.D.
USDA-ARS GPSR
Ft. Collins, CO 80521
(970)490-8345
bai@gpsr.colostate.edu
http://www.gpsr.colostate.edu/GPSR/higraph/people/jaybai.htm

Table One MarcoData

A	504	466	485	465	436	473	516	481	531	620	624	670	639	621
B	280	271	275	267	261	260	265	253	235	232	241	236	241	247
C	1453	1460	1520	1459	1523	1440	1600	1536	1442	1270	1248	1071	1149	1112
D	1344	1368	1352	1361	1429	1415	1388	1347	1387	1406	1438	1425	1454	1417
E	368	374	410	430	430	409	434	406	384	361	388	381	411	404

Table Two MarcoOut (This was a computer output ASCII file. Readers may need
to aligned them to read it.)


K=1            P       D       E     IV        R     T        Pk+1
  ------------------------------------------------------------------
         1  504.00  504.00  504.00  0.2407    0.00  1.0000  504.00
         2  280.00  280.00  280.00  0.1337    0.00  1.0000  280.00
         3 1453.00 1453.00 1453.00  0.6938    0.00  1.0000 1453.00
         4 1344.00 1344.00 1344.00  0.6418    0.00  1.0000 1344.00
         5  368.00  368.00  368.00  0.1757    0.00  1.0000  368.00
V-SUM              2094.13 2094.13                         2094.13
Cos                   0.0000  0.0000
SMC                   0.0000


k=2            P       D       E     IV        R     T        Pk+1
  ------------------------------------------------------------------
         1  504.00  466.00  473.60  0.2252  -30.40  0.9356  436.00
         2  280.00  271.00  272.80  0.1297   -7.20  0.9701  262.89
         3 1453.00 1460.00 1458.60  0.6935    5.60  0.9995 1459.29
         4 1344.00 1368.00 1363.20  0.6481   19.20  1.0099 1381.55
         5  368.00  374.00  372.80  0.1773    4.80  1.0087  377.24
V-SUM              2105.59 2103.22                         2107.07
Cos                   0.9998  0.9999
SMC                   1.0001


k=3            P       D       E     IV        R     T        Pk+1
  ------------------------------------------------------------------
         1  436.00  485.00  475.20  0.2220   39.20  0.9860  478.23
         2  262.89  275.00  272.58  0.1274    9.69  0.9819  270.03
         3 1459.29 1520.00 1507.86  0.7045   48.56  1.0159 1544.18
         4 1381.55 1352.00 1357.91  0.6345  -23.64  0.9789 1323.48
         5  377.24  410.00  403.45  0.1885   26.21  1.0635  436.04
V-SUM              2148.78 2140.20                         2151.25
Cos                   0.9996  0.9995
SMC                   0.9999


k=4            P       D       E     IV        R     T        Pk+1
  ------------------------------------------------------------------
         1  478.23  465.00  467.65  0.2208  -10.58  0.9943  462.35
         2  270.03  267.00  267.61  0.1263   -2.42  0.9919  264.85
         3 1544.18 1459.00 1476.04  0.6968  -68.15  0.9890 1443.02
         4 1323.48 1361.00 1353.50  0.6390   30.01  1.0071 1370.65
         5  436.04  430.00  431.21  0.2036   -4.83  1.0799  464.35
V-SUM              2110.31 2118.23                         2112.00
Cos                   0.9996  0.9995
SMC                   0.9998


k=5            P       D       E     IV        R     T        Pk+1
  ------------------------------------------------------------------
         1  462.35  436.00  441.27  0.2028  -21.08  0.9186  400.53
         2  264.85  261.00  261.77  0.1203   -3.08  0.9523  248.56
         3 1443.02 1523.00 1507.00  0.6926   63.98  0.9940 1513.83
         4 1370.65 1429.00 1417.33  0.6514   46.68  1.0195 1456.82
         5  464.35  430.00  436.87  0.2008  -27.48  0.9863  424.12
V-SUM              2191.96 2175.77                         2194.55
Cos                   0.9997  0.9996
SMC                   0.9999


k=6            P       D       E     IV        R     T        Pk+1
  ------------------------------------------------------------------
         1  400.53  473.00  458.51  0.2140   57.98  1.0554  499.19
         2  248.56  260.00  257.71  0.1203    9.16  1.0000  259.99
         3 1513.83 1440.00 1454.77  0.6791  -59.06  0.9805 1411.92
         4 1456.82 1415.00 1423.36  0.6645  -33.45  1.0200 1443.34
         5  424.12  409.00  412.02  0.1923  -12.10  0.9579  391.80
V-SUM              2129.42 2142.13                         2132.38
Cos                   0.9995  0.9994
SMC                   1.0000


k=7            P       D       E     IV        R     T        Pk+1
  ------------------------------------------------------------------
         1  499.19  516.00  512.64  0.2313   13.45  1.0806  557.61
         2  259.99  265.00  264.00  0.1191    4.01  0.9901  262.38
         3 1411.92 1600.00 1562.38  0.7049  150.47  1.0380 1660.85
         4 1443.34 1388.00 1399.07  0.6313  -44.27  0.9500 1318.64
         5  391.80  434.00  425.56  0.1920   33.76  0.9983  433.25
V-SUM              2238.61 2216.31                         2250.49
Cos                   0.9982  0.9980
SMC                   0.9998


k=8            P       D       E     IV        R     T        Pk+1
  ------------------------------------------------------------------
         1  557.61  481.00  496.32  0.2285  -61.29  0.9880  475.25
         2  262.38  253.00  254.88  0.1174   -7.50  0.9853  249.27
         3 1660.85 1536.00 1560.97  0.7188  -99.88  1.0196 1566.12
         4 1318.64 1347.00 1341.33  0.6176   22.68  0.9784 1317.92
         5  433.25  406.00  411.45  0.1895  -21.80  0.9867  400.60
V-SUM              2152.65 2171.72                         2153.63
Cos                   0.9999  0.9991
SMC                   0.9992


k=9            P       D       E     IV        R     T        Pk+1
  ------------------------------------------------------------------
         1  475.25  531.00  519.85  0.2447   44.60  1.0706  568.50
         2  249.27  235.00  237.85  0.1120  -11.42  0.9539  224.17
         3 1566.12 1442.00 1466.82  0.6904  -99.29  0.9605 1385.07
         4 1317.92 1387.00 1373.18  0.6463   55.26  1.0464 1451.41
         5  400.60  384.00  387.32  0.1823  -13.28  0.9622  369.49
V-SUM              2118.44 2124.63                         2129.55
Cos                   0.9986  0.9984
SMC                   0.9998


k=10           P       D       E     IV        R     T        Pk+1
  ------------------------------------------------------------------
         1  568.50  620.00  609.70  0.2964   41.20  1.2115  751.11
         2  224.17  232.00  230.43  0.1120    6.27  1.0007  232.17
         3 1385.07 1270.00 1293.01  0.6286  -92.05  0.9105 1156.39
         4 1451.41 1406.00 1415.08  0.6880  -36.33  1.0645 1496.63
         5  369.49  361.00  362.70  0.1763   -6.79  0.9673  349.19
V-SUM              2039.19 2056.87                         2077.77
Cos                   0.9963  0.9993
SMC                   1.0030


k=11           P       D       E     IV        R     T        Pk+1
  ------------------------------------------------------------------
         1  751.11  624.00  649.42  0.3155 -101.69  1.0643  664.13
         2  232.17  241.00  239.23  0.1162    7.07  1.0374  250.01
         3 1156.39 1248.00 1229.68  0.5974   73.29  0.9503 1185.94
         4 1496.63 1438.00 1449.73  0.7043  -46.90  1.0237 1472.05
         5  349.19  388.00  380.24  0.1847   31.05  1.0475  406.44
V-SUM              2055.08 2058.49                         2059.65
Cos                   0.9998  0.9978
SMC                   0.9981


k=12           P       D       E     IV        R     T        Pk+1
  ------------------------------------------------------------------
         1  664.13  670.00  668.83  0.3383    4.69  1.0724  718.53
         2  250.01  236.00  238.80  0.1208  -11.21  1.0394  245.31
         3 1185.94 1071.00 1093.99  0.5534  -91.95  0.9264  992.19
         4 1472.05 1425.00 1434.41  0.7256  -37.64  1.0303 1468.21
         5  406.44  381.00  386.09  0.1953  -20.35  1.0573  402.85
V-SUM              1956.38 1976.81                         1969.47
Cos                   0.9970  0.9996
SMC                   1.0025


k=13           P       D       E     IV        R     T        Pk+1
  ------------------------------------------------------------------
         1  718.53  639.00  654.91  0.3264  -63.62  0.9646  616.41
         2  245.31  241.00  241.86  0.1205   -3.45  0.9978  240.46
         3  992.19 1149.00 1117.64  0.5570  125.45  1.0064 1156.40
         4 1468.21 1454.00 1456.84  0.7260  -11.36  1.0006 1454.80
         5  402.85  411.00  409.37  0.2040    6.52  1.0446  429.31
V-SUM              2017.33 2006.61                         2018.86
Cos                   0.9994  0.9976
SMC                   0.9983


k=14           P       D       E     IV        R     T        Pk+1
  ------------------------------------------------------------------
         1  616.41  621.00  620.08  0.3141    3.67  0.9623  597.59
         2  240.46  247.00  245.69  0.1244    5.23  1.0325  255.02
         3 1156.40 1112.00 1120.88  0.5677  -35.52  1.0193 1133.47
         4 1454.80 1417.00 1424.56  0.7215  -30.24  0.9938 1408.26
         5  429.31  404.00  409.06  0.2072  -20.25  1.0156  410.30
V-SUM              1963.24 1974.32                            0.00
Cos                   1.0000  0.9999
SMC                   0.9999



-----Original Message-----
From:	marco [SMTP:sandeco@MBOX.VOL.IT]
Sent:	Tuesday, June 09, 1998 3:23 AM
To:	ECOLOG-L@UMDD.UMD.EDU
Subject:	number of species -Mathem. predetermination-

Hello everyone,
Who could be useful for this problem.
I am a biologist and in my ecology researches I try to forecast what the 
evolution in the number of animal species will be assumed there is an 
interrelation with cohabitant living organisms. Example:
In a pond we find 4 species of living organisms : A, B, C, D, In monthly 
surveys we observe if their number increased, decreased, or remained constant 
thus assessing them respectively 1, 2, X.
After 10 observations the result will be:
1 1 1 2 1 2 1 X X 2   for specie A
2 1 1 1 2 1 1 2 1 2    for            B
1 2 1 X 1 2 2 2 1 1    for            C
2 1 2 1 1 2 1 2 2 2    for            D

What would the next sequence of number will be ?
There always is a law which is non-visible to the eye but identifyable
through a system. Is there a software available able to work this out
?. Thanks. Marco/Italy