#Peat Fire Risk (peatfr) Package

The peat fire risk package or peatfr package provide comprehensive framework to forecast tropical peatland fire through stochastics, machine learning, and optimisation methods. The peatfr consists of three main functions, including, data imputation process, time series forecasting, and drought fire index calculation. This package provides four imputation methods, which can be employed to estimate the missing values of time series data of water table, soil moisture, rainfall, and air temperature. The forecasting process employs the ARIMA stochastic model with Box-Cox transformation, the Long Short-Term Memory (LSTM) method, and the Gated Recurrent Unit (GRU) method. The drought fire index is calculated through Peat Fire Vulnerability Index with Nelder-Mead optimisation to obtain the best parameters automatically.

#Functions

• autopeatfr(WT, SM, Rf, Temp, imputation = c("knn", "spline", "linear", "loess"), model= c("arima", "lstm", "gru"), look_back=12, layer_units=c(32, 32), epochs=100, h = 4, R0 = 3000, dt = 1)
• knn_imputation(WT, SM, Rf, Temp, k = 5)
• spline_interpolation(WT, SM, Rf, Temp)
• loess_interpolation(WT, SM, Rf, Temp, span = 0.5)
• linear_interpolation(WT, SM, Rf, Temp)
• autopredictarima(WT, SM, Rf, Temp, h, test_split_ratio = 0.2)
• autopredictlstm(WT, SM, Rf, Temp, h, look_back =12, lstm_layers_unit = c(50), epochs = 100)
• autopredictgru(WT, SM, Rf, Temp, h, look_back =12, lstm_layers_unit = c(50), epochs = 100)
• firepredict(WT, SM, Rf, Temp, R0 = 3000, dt = 1, h)

#Getting Started

#Prerequisites

1. VIM package
2. stats package
3. forecast package
4. ggplot2 package

Install the dependencies using:

install.packages("devtools") # If not installed yet
devtools::install_github("mellygsln/peatfr", dependencies = TRUE)

library(peatfr)

#Project Structure

peatfr/
|-- man/                     
    |-- peatfirerisk.Rd/
|-- R                
    |-- Function Auto Peat Fire Risk/
    |-- Function Auto Predict ARIMA/
    |-- Function Auto Predict LSTM/
    |-- Function Auto Predict GRU/
    |-- Function Fire Predict/
    |-- Function KNN/
    |-- Function Linear/
    |-- Function Loess/
    |-- Function Spline/
|-- .Rbuildignore                
|-- .Rhistory                        
|-- DESCRIPTION
|-- NAMESPACE
|-- peatfirerisk.Rproj