20 de janeiro de 2015

Introduction, Dataset and Model

Demand planning and forecasting is a business process that involves predicting future demand for products and services. The main goal at this project is create a model using ARIMA technique for forecast a demand of a product sale in a supermarket store in Brazil. This product is a 250g pack of coffee premium.

  • The dataset are real information about sale of product in a supermarket (only at one store) in Brazil
  • This dataset contain data since 2009, but for this presentation we are going to use only 2014 and beginning of 2015.
  • This is URL: https://lcavalcanti.shinyapps.io/Cafe/ of a shiny app developed for this project.
  • We use forecast library for this project.
  • Forecasting methodology used is classic ARIMA.

Create the Model for Forecast

Cafe = read.csv("Coffee.csv")
Cafe$Date = as.Date(Cafe$Date, format = "%d/%m/%Y")
fit = auto.arima(Cafe$Qnt)
fcast = forecast(fit)
fcast
##     Point Forecast    Lo 80     Hi 80     Lo 95    Hi 95
## 379       52.37993 15.19366  89.56620 -4.491564 109.2514
## 380       50.62175 12.77535  88.46814 -7.259322 108.5028
## 381       53.56832 14.21273  92.92392 -6.620866 113.7575
## 382       56.69987 16.21961  97.18013 -5.209347 118.6091
## 383       59.17881 17.98549 100.37212 -3.820931 122.1785
## 384       60.78505 19.16075 102.40934 -2.873822 124.4439
## 385       61.65533 19.75439 103.55627 -2.426629 125.7373
## 386       62.02549 19.92262 104.12837 -2.365301 126.4163
## 387       62.11097 19.83818 104.38376 -2.539686 126.7616
## 388       62.06560 19.63347 104.49773 -2.828741 126.9599

Plot with forecast

plot(fcast, main = "Quantity coeffe", 
     ylab = "Daily Sale of Coffee in Quantity", xlab = "Days")
grid()

Conclusion

The developed forecasting models leave a considerable amount of valuable information for the business.We decide create a model using ARIMA with some dataset provided by supermarket chain. This dataset has information since 2009 until begin of 2015. Better short-term forecasts allow the supermarket chain to reduce inventory costs and improve their operation margins, in my opinion a very challenging topic in this industry.