The included dataset, broken down over the next 8 slides, comes from the Rfssa library, and records the calls received at an anonymous bank’s call center for the year 1999.

The original dataset used the following variables, which you’ll see in the various graphs:

calls - number of calls received, typically summarized into a mean or sum total

u - a unit of time lasting 6 minutes, starting at 0000 and ending at 2359

Date - the full calendar date on which a call was received (yyyy-mm-dd)

Day - The day of the week on which a call was received

Month - The calendar month on which a call was received, abbreviated to the first three letters of the month

fig1<-ggplot(data= Call_data_week,aes(x=Day, y=numCalls))+
  geom_bar(stat="identity", width = 0.8, 
  fill=c("red","orange","yellow","green","blue","blueviolet","violet"))+
  geom_text(aes(label=round(numCalls, 2)),nudge_y = 2000)+
  labs(y="Total Number of Calls on that day")
fig1

(Using the Data from the previous plot)

Sunday + Monday + Tuesday + Wednesday + Thursday + Friday + Saturday = ??

81038 + 82772 + 77836 + 79727 + 28548 + 10736 + 84757.5 = 445414.5

\({445414.5 \over 365} \approx 1220.314\) calls received per day

• Friday and Saturday typically see the lowest call volume for the week
• Highest call volume is observed on Sundays, with Monday and Tuesday usually close behind
• Call volume is consistently low outside of ‘normal’ business hours (u=80 to u=180)
• Call volume is at it’s peak around 10:15 ( \({103 * 6 \over 60} = 10.3\) hours), and stays consistently high through typical lunch times.

fig3<-ggplot(data= Call_data_month,aes(x=Month, y=numCalls))+
  geom_bar(stat="identity", width = 0.8, 
  fill=c("red","orangered","orange",  "gold", "yellow", "yellowgreen", 
         "green", "cyan3", "blue", "purple2", "blueviolet", "violet"))+
  geom_text(aes(label=round(numCalls, 2)),nudge_y = 2000)+
  labs(y="Total Number of Calls on that day")
fig3

• Call volume per day does not seem to be significantly impacted by Federal or ‘bank’ holidays
  • Thanksgiving Day (Day 329), and Christmas Day (Day 359), each seem to be local minima, but not by significant margins
  • Memorial Day (Day 141), July 24 (Day 205), and September 11 (Day 254) are the three lowest volume days of the year, but only Memorial Day held any special significance in 1999
• Spikes in calls per day appear to be more significantly influenced by time of month
• Another notable trend is that call volume tends to increase in the final month of a fiscal quarter
  • September is a notable exception, as August has the relative spike for its quarter

Overall, this call center appeared to experience two heavy volume increases, once in August, and again in November, sustained through December. Notably, November and December would have been just before the Y2K scare, which may have concerned users of the bank and led them to call in more.

Furthermore, the bank’s customers seemed to call in the most on Sunday, and the least on Friday and Saturday.

This all means that the optimal times to call (to minimize potential wait times) would be the ‘off-hours’ (before 6am and after 6pm) on a Friday or Saturday, during either the second or first month of a fiscal quarter.