Readme.Rmd

---
output: github_document
---

<!-- README.md is generated from README.Rmd. Please edit that file 
To regenerate, run `quarto::quarto_render("Readme.Rmd", output_file = "README.md")`
-->

```{r, include = FALSE}
knitr::opts_chunk$set(
  collapse = TRUE,
  comment = "#>",
  fig.path = "man/figures/README-",
  out.width = "100%"
)
```

# rabbit

<!-- badges: start -->
<!-- badges: end -->

The goal of rabbit is to ...

Use package `RcppRoll` to optimise claulcations of metrics clculated in wolling windows, e.g. rolling mean, var, sd etc.

## Installation

You can install the development version of rabbit from [GitHub](https://github.com/) with:

```{r install, eval=FALSE}
# install.packages("devtools")
 devtools::install_github("traitecoevo/rabbit")
```

## Example

This is a basic example:

```{r example}
library(rabbit)
library(dplyr)
```
super-fast version of the rolling window calculations:

example file is less than one hour of a bilby called piccolo:
```{r}
df <- arrow::read_parquet("tests/testthat/raw_Pic2Jan_50000.parquet")
nrow(df)
```
the new version is pretty fast:
```{r}
system.time(dat <- moving_window_calcs_2(df))
```

## Identifying high sumVDBA times

sumVDBA is the best measure we have of heat-generating movement or activities:

```{r,warning=FALSE}
library(ggplot2)
dat %>%
  ggplot(aes(x = time, y = sumVDBA)) +
  geom_point(size = 0.2) + theme_classic()
```

Now can classify all these movements based on a pre-built classifier from a zoo animal:

```{r,warning=FALSE}
# load a classifcation object
MSOM = readRDS("tests/testthat/MSOM_8by7_small.rds")

# make predictions
nighttime_activities <- classify_behaviors(dat, MSOM)
```

we can order the activities by their estimated energy use / heat production:

```{r}
nighttime_activities <- 
  nighttime_activities %>%
  filter(!is.na(behavior)&!is.na(sumVDBA)) %>%
  mutate(behavior = forcats::fct_reorder(behavior, sumVDBA, .fun = median, na.rm = TRUE))
```

and we can plot energy use and estimated activity through time
```{r}
nighttime_activities %>%
  ggplot(aes(x = time, y = sumVDBA, col = behavior)) +
  geom_point(alpha = 0.7, size = 0.2) + theme_classic()
```

the data seems to at a temporal resolution of 25 readings per second which might be a lot to handle for various graphing applications.  Here is a function to take the mode of the categorical variables at a given resolution.

```{r}
window_in_minutes <- 1

minute_summary <- summary_by_time(nighttime_activities,
                                      time_col = "time",
                                      behavior_col = "behavior",
                                      window_minutes=window_in_minutes
                                      )

minute_summary %>%
ggplot(aes(x = window, y = mode_behavior, color = mode_behavior)) +
    geom_point(size = 3) +
    labs(
      x = "Time",
      y = "Mode Behavior",
      title = paste0("Most Common Behavior Within Every ", window_in_minutes, " Minute(s)")
    ) +
    theme_minimal() -> plotting
  print(plotting)
```