please help me in interpretation

2620 views
Skip to first unread message

Kamakshaiah

unread,
Oct 5, 2011, 3:50:03 AM10/5/11
to FactoMineR users
Sir,

how to interpret this output with respect to column headings, i.e.,
Cla/Mod, Mod/Cla, Global, p-value, v.test...

Mod/Cla (even though, I don't know what it mean), I understood it
explains that particular variable (sugar=no sugar) as 81% of
respondents take tea with no sugar. the same is true only 51% to the
population. Regarding p-value, all of them are below 5% does it mean
that all variables contributes significantly. and what is this v.test?
Please help me how to interpret these variables with respect to column
headings.

res.hcpc$desc.var$category
$`1`
Cla/
Mod Mod/Cla Global p.value v.test
sugar=no sugar
50.322581 81.250000 51.66667 1.207794e-12 7.104477
sex=F
46.067416 85.416667 59.33333 1.138208e-10 6.447353
sport=no sportive
52.066116 65.625000 40.33333 2.041112e-09 5.994501
feminine=feminine
48.837209 65.625000 43.00000 1.076599e-07 5.313296
socio.professional.category=non active
60.937500 40.625000 21.33333 1.188542e-07 5.295249
age=60 and more
71.052632 28.125000 12.66667 2.675958e-07 5.144947
tea.type=black
52.702703 40.625000 24.66667 3.103710e-05 4.165720
afternoon.tea=afternoon tea
42.011834 73.958333 56.33333 3.152669e-05 4.162149
location.of.purchase=large-scale retail stores+specialized shop
51.282051 41.666667 26.00000 5.716069e-05 4.024236
age=45-59
54.098361 34.375000 20.33333 9.940768e-05 3.892033
diuretic=diuretic
40.229885 72.916667 58.00000 4.429275e-04 3.513090
age=35-44
57.500000 23.958333 13.33333 6.196066e-04 3.422883
healthy=healthy
38.095238 83.333333 70.00000 6.276591e-04 3.419372
refined=refined
37.674419 84.375000 71.66667 9.261828e-04 3.312040
tea.house=tea house
50.000000 30.208333 19.33333 2.308576e-03 3.047364
after.dinner=no after dinner
34.050179 98.958333 93.00000 5.124549e-03 2.799099
slimming=slimming
51.111111 23.958333 15.00000 6.145008e-03 2.739942
frequency=more than 2/day
40.944882 54.166667 42.33333 6.653848e-03 2.713690
conviviality=conviviality
35.537190 89.583333 80.66667 8.946716e-03 2.614084
how=other
77.777778 7.291667 3.00000 1.145195e-02 2.528597
evening=no evening
36.548223 75.000000 65.66667 2.573663e-02 2.230163
relaxing=no relaxing
39.823009 46.875000 37.66667 3.393732e-02 2.120816
work=work
41.379310 37.500000 29.00000 3.832284e-02 2.071385
shape=bulk
16.666667 6.250000 12.00000 4.806758e-02 -1.976770
work=no work
28.169014 62.500000 71.00000 3.832284e-02 -2.071385
relaxing=relaxing
27.272727 53.125000 62.33333 3.393732e-02 -2.120816
evening=evening
23.300971 25.000000 34.33333 2.573663e-02 -2.230163
frequency=1 to 2/week
15.909091 7.291667 14.66667 1.665451e-02 -2.394247
conviviality=no conviviality
17.241379 10.416667 19.33333 8.946716e-03 -2.614084
slimming=no slimming
28.627451 76.041667 85.00000 6.145008e-03 -2.739942
after.dinner=after dinner
4.761905 1.041667 7.00000 5.124549e-03 -2.799099
tea.house=no tea house
27.685950 69.791667 80.66667 2.308576e-03 -3.047364
refined=no refined
17.647059 15.625000 28.33333 9.261828e-04 -3.312040
healthy=no healthy
17.777778 16.666667 30.00000 6.276591e-04 -3.419372
diuretic=no diuretic
20.634921 27.083333 42.00000 4.429275e-04 -3.513090
tea.type=perfumed
24.352332 48.958333 64.33333 2.591569e-04 -3.653037
location.of.purchase=specialized shop
3.333333 1.041667 10.00000 1.568457e-04 -3.779973
afternoon.tea=no afternoon tea
19.083969 26.041667 43.66667 3.152669e-05 -4.162149
age=25-34
11.594203 8.333333 23.00000 2.250914e-05 -4.238427
feminine=no feminine
19.298246 34.375000 57.00000 1.076599e-07 -5.313296
sport=sportive
18.435754 34.375000 59.66667 2.041112e-09 -5.994501
sex=M
11.475410 14.583333 40.66667 1.138208e-10 -6.447353
age=15-24
5.434783 5.208333 30.66667 2.043546e-12 -7.031480
sugar=sugar
12.413793 18.750000 48.33333 1.207794e-12 -7.104477
socio.professional.category=student
0.000000 0.000000 23.33333 3.373472e-14 -7.583118

François Husson

unread,
Oct 6, 2011, 8:55:55 AM10/6/11
to factomin...@googlegroups.com
If I interpret the first line for the description of class 1:

Class 1:


Cla/ Mod   Mod/Cla   Global      p.value    v.test
sugar=no sugar
50.322581 81.250000 51.66667 1.207794e-12  7.104477

Cla/Mod: 50.3% of the individuals who take no sugar also belongs to class 1
Mod/Cla: 81.5% of individuals of class 1 do no take sugar
Global: 51.67% of individuals do not take sugar.

All the categories that are in this output are the significantly ones. You can modify the 5% threshold if you want using the proba argument.

kamakshaiah m

unread,
Oct 7, 2011, 12:05:27 AM10/7/11
to factomin...@googlegroups.com
Sir,

Thanks,
In deed, I am more concerned about the last two values, i.e., p-value and v-test.

p-value is less than 0.05, so, shall I interpret that the difference is significant, which means, all the categories are significantly different. and regarding v-test, usually, in statistics, 'v' is used in kuiper test (modified form of KS test), to test distribution. Is it the same is used here? v-test here, does it denotes 'v statistic' of kuiper test? Otherwise, how should I explain this value (v-test).

Please help me...
--
-----
Kamakshaiah Musunuru
+251-914238020 (Ethiopia)

François Husson

unread,
Oct 7, 2011, 11:35:07 AM10/7/11
to factomin...@googlegroups.com
The p-value is less than 5% when one category is significantly linked to another categories. If the v-test is positive, it means that the category is over-expressed for the category and if the v-test is negative it means that the category is under-expressed for the category.

For example, the modality no sugar is significantly more present in the class 1 than in the other classes.
And, if you see the last line of the table, you can see that the v-test for sugar is negative, which means that the modality sugar is significantly less present in class 1 (what is expected! but when one variable has more than 2 categories, it is not symmetric).

The v-test is just a transformation of the p-value in a quantile of the normal law. What is only interesting in this v-test is only the sign of the v-test.

Kamakshaiah

unread,
Nov 10, 2011, 1:16:43 AM11/10/11
to FactoMineR users
Thanks and dear sir,
in our previous conversation you said like below:

Cla/Mod: 50.3% of the individuals who take no sugar also belongs to
class 1
Mod/Cla: 81.5% of individuals of class 1 do no take sugar
Global: 51.67% of individuals do not take sugar.

it is little confusing to me (pardon for my being little dull in
comprehension). The first two statements are alike. Both (50.3% and
81.5%) do not take sugar as it seems to me, then where is the
difference in Cla/Mod & Mod/Cla. Is it that 50.3% is from population,
if it is true, then how to distinguish it (Cla/Mod) from Global.

Regards,
On Oct 7, 8:35 pm, François Husson <francois.hus...@agrocampus-

Kamakshaiah

unread,
Nov 10, 2011, 1:21:28 AM11/10/11
to FactoMineR users
Sir,
If I am correct, you said, if p-value is less than 5% then the link is
significant, in this example (tea example) all p-values are less than
5%, does it mean that all the variables link significantly with
respective v-test values.

Please let me know

Regards,

On Oct 7, 8:35 pm, François Husson <francois.hus...@agrocampus-
ouest.fr> wrote:

François Husson

unread,
Nov 10, 2011, 3:41:01 AM11/10/11
to factomin...@googlegroups.com
Only the variables and categories with a p-value less than 5% are given, that is why in the outputs all the p-values are small. But there are many more categories than the ones given in the output.
FH

François Husson

unread,
Nov 10, 2011, 3:44:52 AM11/10/11
to factomin...@googlegroups.com
The first two statements are not alike. The first one is focused on the understanding of people who do not take sugar: 50.3% of them belongs to class 1. The second one is focused on the undestanding of class 1:  81.5% of individuals of class 1 do no take suga. It is not the same information.
And then, global corresponds to the percentage of individuals who do not take sugar in the population.

FH

Kamakshaiah

unread,
Nov 10, 2011, 11:51:25 PM11/10/11
to FactoMineR users
Thanks..I understood this with the help of your book. Although I did
not understand the concept fully, now I am little more confident than
before to help other here.

Thanks again....

On Nov 10, 1:44 pm, François Husson <francois.hus...@agrocampus-

philippe

unread,
Nov 20, 2011, 9:18:47 AM11/20/11
to factomin...@googlegroups.com
j’ai retrouvé où ça bloque mais cela ne  relève pas de factominer ?
dans la version  bits de R
il n’est pas possible d’importer directement des fichiers Excel
 
The following object(s) are masked from 'package:Rcmdr':
 
    defmacro
 
Le chargement a nécessité le package : RODBC
Error in odbcConnectExcel(File) :
  odbcConnectExcel is only usable with 32-bit Windows
 
voici l’erreur
Philippe Gaillard
 
 
 
 
Reply all
Reply to author
Forward
0 new messages