index.Rmd

---
title: "Portal Forecast"
---

```{r setup, include=FALSE}
knitr::opts_chunk$set(echo = TRUE)
library(dplyr)
library(portalr)
source("forecast_tools.R")
```

## Total Abundance Forecast

This is the forecast for next month's sampling of rodents at Portal.

```{r, echo=FALSE, message=FALSE, warning=FALSE}
obs_data = portalr::abundance("repo")
obs_data$total = rowSums(select(obs_data, -period))
new_moons = read.csv(text = RCurl::getURL("https://raw.githubusercontent.com/weecology/PortalData/master/Rodents/moon_dates.csv"))

obs_data_newmoon = inner_join(obs_data, new_moons, by = c("period" = "period"))
obs_data_newmoon$censusdate = as.Date(obs_data_newmoon$censusdate)
#for_data = read.csv("predictions/2017-02-24Allforecasts.csv")

for_data = compile_forecasts() %>%
  dplyr::filter(level == 'All', model == 'Ensemble', date == max(date))

most_recent_forecast = max(for_data$date)
for_data = filter(for_data, date == most_recent_forecast)

for_data$model = "ensemble"
for_data$forecastdate = as.Date(paste(for_data$forecastyear,for_data$forecastmonth,'15',sep='-'))

forecast_viz(obs_data = obs_data_newmoon,
             obs_date_col_name = "censusdate",
             obs_val_col_name = "total",
             for_data = for_data,
             for_date_col_name = "forecastdate",
             for_val_col_name = "estimate",
             for_model_name = "ensemble",
             for_lowerpi_col_name = "LowerPI",
             for_upperpi_col_name = "UpperPI",
             start_newmoon = 300,
             ylabel = 'Total Abundance')
```

## Species-Level Forecasts

```{r speciesforecasts, echo=FALSE, message=FALSE, warning=FALSE}
source("forecast_tools.R")

data = compile_forecasts()
ensemble = dplyr::filter(data, level == 'All', model == 'Ensemble', date == max(date))
sp_predictions = get_sp_predicts(ensemble, 'All', lead_time = 1)
sp_predict = plot_species_forecast(sp_predictions)
plot(sp_predict)
```
```{r highabund_ts_forecasts, echo=FALSE, message=FALSE, warning=FALSE}
most_abund_sp = sp_predictions %>% 
  filter(species != 'total') %>% 
  arrange(desc(estimate)) %>% 
  head(3) %>% 
  select(species)

# load in rodent species table to get scientific names to display on plots
species_table = read.csv(
  text = RCurl::getURL(
      "https://raw.githubusercontent.com/weecology/PortalData/master/Rodents/Portal_rodent_species.csv"),stringsAsFactors = F,na.strings = '')
most_abund_sp_names = filter(species_table,speciescode %in% most_abund_sp$species) %>% select(speciescode,scientificname)

for (n in seq(dim(most_abund_sp_names)[1])) {
  species_forecast = forecast_viz(obs_data = obs_data_newmoon,
               obs_date_col_name = "censusdate",
               obs_val_col_name = most_abund_sp_names$speciescode[n],
               for_data = for_data,
               for_date_col_name = "forecastdate",
               for_val_col_name = "estimate",
               for_model_name = "ensemble",
               for_lowerpi_col_name = "LowerPI",
               for_upperpi_col_name = "UpperPI",
               start_newmoon = 300,
               ylabel = most_abund_sp_names$scientificname[n])
  plot(species_forecast)
}
```