In order to build a predictive model, we need to prepare our data first, so that we can feed in the relevant data into the model. Take a look at the following code:
# group data into every 3 monthslibrary(lubridate)ordersDF$Quarter = as.character(round_date(ordersDF$InvoiceDate, '3 months'))dataDF <- ordersDF %>% group_by(CustomerID, Quarter) %>% summarize(SalesSum=sum(Sales), SalesAvg=mean(Sales), SalesCount=n())
As you can see from this code, we are using the lubridate package that is going to help us to handle data with dates more easily. Using the round_date function in the lubridate package, we first round InvoiceDate to the nearest quarter. Then, we group the data by CustomerID and the newly-created column, Quarter ...
