Description
'M5 Forecasting' Challenges Data.
Description
Contains functions, which facilitate downloading, loading and preparing data from 'M5 Forecasting' challenges (by 'University of Nicosia', hosted on 'Kaggle'). The data itself is set of time series of different product sales in 'Walmart'. The package also includes a ready-to-use built-in M5 subset named 'tiny_m5'. For detailed information about the challenges, see: Makridakis, S. & Spiliotis, E. & Assimakopoulos, V. (2020). <doi:10.1016/j.ijforecast.2021.10.009>.
README.md
m5 
M5 Walmart Challenge Data
Installation
You can install the development version of m5 from GitHub with:
# install.packages("devtools")
devtools::install_github("krzjoa/m5")
Usage
library(m5)
library(zeallot)
library(ggplot2)
DIR <- 'data'
# Downloading the data
m5_download(DIR)
# Loading the data
c(sales_train,
sales_test,
sell_prices,
calendar,
weights) %<-% m5_get_raw_evaluation(DIR)
# Preparing the data
m5_data <-
m5_prepare(sales_train, sales_test, calendar, sell_prices)
# Demand classification
m5_demand <- m5_demand_type(m5_data)
foods_1_demand <-
m5_demand[startsWith(as.character(m5_demand$item_id), "FOODS_1")]
plot <-
ggplot(foods_1_demand) +
geom_point(aes(log(cv2), log(adi),
item_id = item_id, col = demand_type)) +
geom_hline(yintercept = log(1.32)) +
geom_vline(xintercept = log(0.49)) +
theme_minimal()
plot
