randomForestExample/chase_experiment.Rmd at master · messDiv/randomForestExample · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
---
title: "Chase_trial"
output: html_document
---

```{r setup, include=FALSE}
knitr::opts_chunk$set(echo = TRUE)
require("mobsim")
require("vegan")
require("randomForest")
```

## R Markdown

Exploring how all of this works.

```{r}

# Species abundances
sad1 <- sim_sad(100, 2000, sad_type = "lnorm", sad_coef = list("meanlog" = 2, "sdlog" = 1))
rc1 <- rare_curve(sad1)
plot(rc1, type = "l", xlab = "Sample size", ylab = "Expected species richness")
renyi(sad1)

# Spatial distributions
comm_rand <- sim_poisson_community(s_pool = 30, n_sim = 300)
comm_agg <- sim_thomas_community(s_pool = 30, n_sim = 300, sigma = 0.05, mother_points = 1)
par(mfrow = c(1,2))
plot(comm_rand)
plot(comm_agg)

sar_rand <- divar(comm_rand)
sar_agg <- divar(comm_agg)

plot(m_species ~ prop_area, data = sar_rand, type = "b", log = "xy",
     xlab = "Proportion of area sampled",ylab = "No. of species",
     ylim = c(3,30))
lines(m_species ~ prop_area, data = sar_agg, type = "b", col = 2)
legend("bottomright", c("Random","Aggregated"), col = 1:2, lwd = 2)


```

Now let's try something dumb

```{r}
foo <- function(x) sim_sad(100, 2000, sad_type = "lnorm", sad_coef = list("meanlog" = 2, "sdlog" = 1))

nrep <- 100
res_lnorm<-matrix(ncol=11, nrow=nrep)
mean_draw <- rlnorm(nrep, meanlog=log(2), sdlog=0.1)
sd_draw <- rlnorm(nrep, meanlog=log(1), sdlog=0.1)

for(i in 1:nrep) {
  ss<-sim_sad(100, 2000, sad_type = "lnorm", sad_coef = list("meanlog" = mean_draw[i], "sdlog" = sd_draw[i]))
  res_lnorm[i,]<-renyi(ss)
}


res_rls<-matrix(ncol=11, nrow=nrep)
N_draw <- rlnorm(nrep, meanlog=log(100), sdlog=0.1)
alpha_draw <- rlnorm(nrep, meanlog=log(100), sdlog=0.1)

for(i in 1:nrep) {
  ss<-sim_sad(s_pool=100, n_sim=2000, sad_type = "ls", sad_coef = list("N" = N_draw[i], "alpha" = alpha_draw[i]))
  res_rls[i,]<-renyi(ss)
}

full<-rbind(res_lnorm, res_rls)
model <- as.factor(rep(c("lnorm", "ls"), each=nrep))

plot(full[1,], type="l", ylim=range(full), col="black", lwd=2)
for(i in 2:100)
  lines(full[i,], col="black", lwd=2)

for(i in 101:200)
  lines(full[i,], col="red", lwd=2)

# random forests
randomForest(full, model)

```

Neg binom and fisher log-series
Set scale k to zero and they converge
pica package on andy's github