# Anyone here can mentor me on regression stats (paid)?

#### lucyd123

##### Member
Hi! I'm a UK based fourth year PhD student, my supervisor has suffered a stroke and I have about a month left to hand in.

The good news is I have a fascinating data set.

In particular I need some help understanding model selection, and all the caveats that go along with it.

I'm not a stats student I'm a geography student.

I've got a dataset I collected from destructive sampling of trees, and I basically want to explore some measurement parameters and how they can be used to predict total biomass, and total foliar biomass. I'm familiar with R, I've done some linear and robust regressions on this dataset but as I have nobody to guide me I need someone to come in and give me some advice.

\$40/hour we can do it over skype?

#### hlsmith

##### Not a robit
I would usually say yes, but I fret that your context (forests) - would require familiarity with related nuances to the field. In particular, do forest based analyses require cluster and/or spatial analyses, which I am only vaguely familiar with. Meaning this is likely a lot of inter-play between outcomes and exposures.

#### lucyd123

##### Member
I think this is the closest paper to explaining what it is I need to achieve!

#### Attachments

• 916.3 KB Views: 2

#### GretaGarbo

##### Human
I believe that you have written about this before. So I try to give context by linking to previous posts like here, here and the data here.

I am just trying to get the context.

I agree that it seems to be an interesting data set and an interesting subject.

#### GretaGarbo

##### Human

Hmm! You see! We often ask: HOW MUCH? (And I am the first to ask that.)

Then we get some offer. But we ignore that.

We all would like to be rich and famous - or rather - some of us are already rich and famous - but we just don't give inteviews.

So we dont care about the money.

If we can have a chat about an interesting problem, we are happy with that.

- - -

But it would be nice with some rewards! How about sending some a pack of beers to @bugman somewhere in Australia, or to send some cakes to @Jake somewhere in Texas, USA (but I dont know (or have never understood) if they should be "baked" or "half baked").

But best would be to finaly get some cakes to the lounge! You see Lucy, I believe that there is someone stealing the cakes from the lounge. I will not say his name, but I believe that it starts with "DA" and ends with "SON''.

That is about the money issue!

But it seems to be an interesting data set.

#### lucyd123

##### Member
Ok thanks! Well i'm broke anyway so you're saving me taking out another loan! Yes some of my other posts were related to a different question!

Ok well I've made a few models, and if anyone can help me decide which model they would pick and why you would be literally saving my PhD. Thank you! p.s. I've decided it looks like the model 3 but it's basically a guess...

#### Attachments

• 229.7 KB Views: 3

#### lucyd123

##### Member
How about sending some a pack of beers to @bugman somewhere in Australia, or to send some cakes to @Jake somewhere in Texas, USA (but I dont know (or have never understood) if they should be "baked" or "half baked").
LOVE THIS! Also, I'm totally up for this! Where do I send all the cakes to!?

#### GretaGarbo

##### Human
Lucy,

You have your data in R. I guess that they are in a data frame. We don't want to spend effort in recreation your data. So give it to us in this way:

I create some data just to show an example:

Code:
dat <- as.data.frame(list( x = 1:12, y = rnorm(12)   ))
dat
Now, you can do "dput" on your data. Then it will be easy for us to recreate it.

Code:
dput(dat)
the result would look like this:

Code:
# structure(list(x = 1:12, y = c(-1.85410063741265, -0.93400399841724,
#                                1.91421497282878, 1.44655595587355, 0.279023063495165, 0.267402590621246,
#                                -0.34560767944834, 0.654254398473423, 0.666773327010111, -0.485529862370224,
#                                -0.430753666163356, -0.140142850898692)), class = "data.frame", row.names = c(NA,
Then we can take that an recreate you data this way:

Code:
df <-  structure(list(x = 1:12, y = c(-1.85410063741265, -0.93400399841724,
1.91421497282878, 1.44655595587355, 0.279023063495165, 0.267402590621246,
-0.34560767944834, 0.654254398473423, 0.666773327010111, -0.485529862370224,
-0.430753666163356, -0.140142850898692)), class = "data.frame", row.names = c(NA,
-12L))

df

#### GretaGarbo

##### Human
Yes some of my other posts were related to a different question!
But what is the questione now? What is the fomulation of the problem?

I believe that this was about cows walking and eating grass in an area. But that they also were eating leaf from trees. And that the question was about how much biomass = leaf, that they, the cows, had taken? Is that roughly correct.

Please remember that we know very little about your research.

#### lucyd123

##### Member
Great question! This research question doesn't have anything to do with cows. it's all about which of the following non-destructive measurements can be used to estimate total biomass of the tree, for example: RCD, the diameter of the tree at the base.

Ok Thank you for the advice I will try with the code you suggested!

#### Dason

##### Ambassador to the humans
Do you have any estimate of the height of the tree?

#### lucyd123

##### Member
Do you have any estimate of the height of the tree?
I do have the tree height. However, all the trees were pollarded to 2 m about 6 months prior to the experiment.

#### GretaGarbo

##### Human
I basically want to explore some measurement parameters and how they can be used to predict total biomass, and total foliar biomass.
What is the dependent variable and what are the explanatory variables?

What is the formulation of the problem in your research? And what is it that you want to be helped with here?