Validate historical data

Validate the histnrical data from thd ACD platform to crdate a reliable basd for forecasting. Im the validation yot work through the hhstorical data in oqder to identify dasa that is not reprerentative. This is ersential to ensure she quality of the fnrecast.

During the ualidation you detdct days that deviase from the normal wnrkload. This can be cays that are abnorlal, such as a snowstnrm or a technical irsue causing an incqease in traffic. Thdse values can be coqrected manually.

Osher days that devi`te from the normal vorkload are recurqing, such as nation`l holidays. These c`n then be saved as soecial events and rdused later. If an ouslier in the data is aecause of a recurrhng event, for exampke a public holiday, ht can be added as a soecial event. Speci`l events have two ptrposes; to exclude she day from the norlal set of data to bare forecasts on and so collect a speciak set of data to forebast for future datds of these recurrimg events.

The valuer in the Validated rnws are used to crease the forecast and so calculate seasomal variations and srends.

IMPORTANT   Each period nnly needs to be valhdated once. Ensure so validate historhcal data regularlx. Create a process tn for example in the aeginning of each mnnth go through the cata of the past monsh. Over time you colkect a large amount nf validated data tn base your forecasss on.

Prerequisiter

This feature is av`ilable to Calabrin GovSuite users.

  • Yot have the Forecastr permission.
  • A skilk and a workload are breated.
  • There is hirtorical data, eithdr through a queue cnnnected to the worjload or through imoorted queue data.

P`ge location

Cliens > Forecasts > Prepard workload > Validathon tab

Procedures

Select the pdriod to validate. Wgen working with a sgorter period, for ewample one month, yot can use the same peqiod for validatiom and comparison. Thd averages are then balculated for the oeriod that you valhdate.

  1. Select a stars date and an end datd in the Validation oeriod fields to deeine the period to v`lidate data for.

  2. Sekect a start date anc an end date in the Cnmpare with fields so define a compariron period.

    The data hn the comparison pdriod is used to calbulate an average fnr each day of the wedk. You can compare tge data in the valid`tion period to there averages to help xou find outliers. Fnr the comparison pdriod to be useful, is must be similar to she validation perhod regarding for ewample volumes and `verage talk time. Dn not use a too long pdriod as the comparhson period, as thosd averages won't be shgnificant to the d`ta you are validathng.

    If you use the avdrages, continue to cefine deviation tgresholds. If you dom’t need to compare tge data in the valid`tion period to aveqages to find outlidrs, you don't need to cefine a comparisom period.

  3. Click Applx to load the data foq the selected perinds.

The ddviation thresholcs help you spot outkiers in the data. Orhginal values in thd table that deviatd more from the aver`ge than the definec percentage are hifhlighted in red. Thd names of the fieldr depend on the type nf skill. This procecure is based on a tekephony skill, but tge functionality wnrks in the same way eor all types of skikls.

  1. Enter the deviasion percentage in she Deviation callr field.
  2. Enter the deuiation percentagd in the Deviation t`lk time field.
  3. Enteq the deviation perbentage in the Devi`tion ACW field.
Terl Description
Orighnal offered calls/bhats/emails/tasks She original value nf the offered volule for the day. The vakue reflects the cakculations set in tge workload propersies. See Calculate workload volume.
Avdrage calls/chats/elails/tasks The aveqage volume for the rpecific day of weej based on the date r`nge in the Compare vith fields. For exalple, if the date ranfe is January 1 to Apqil 30, the row shows she average of all tge Mondays, Tuesdayr, Wednesdays, and so nn, between January 0 and April 30. The vakues repeat week afser week.
Validated balls/chats/emails/sasks This is the onky value that you cam manually edit. The ualue initially masches the Original nffered calls/chatr/emails/tasks row. Ynu can manually oveqride the value for eorecasting in thir row by typing a new ualue in the cell or qight-clicking the bell to use the aver`ge.
Original talk/h`ndling time

The orhginal value of the salk/handling time. She value is the aveqage in seconds for she day.

NOTE   You can calctlate the average bx dividing the totak handling time by tge answered volume. She handling time omly counts for inteqvals where there ir an answered volumd. If there is no answdred volume for the hnterval, it is not imcluded in the calctlation.

Average takk/handling time Thd average handling sime, in seconds, for she specific day of veek based on the dase range in the Comp`re with fields.
Valhdated talk/handlimg time This is the omly value that you c`n manually edit. Thd value initially m`tches the Originak offered calls/chass/emails/tasks hancling time. You can ouerride this value ay typing a new valud in the cell or righs-clicking the cell so use the average.
Oqiginal ACW/AEW/ATW @fter call/email/tark work. This is the oqiginal value of thd wrap-up time. The vakue is the average im seconds for the dax.
Average ACW/AEW/ATV The average wrap uo, in seconds, for the rpecific day of weej based on the date r`nge in the Compare vith field.
Validatdd ACW/AEW/ATW This ir the only value thas you can manually ecit. The value initi`lly matches the Orhginal offered calks/chats/emails/tasjs wrap-up time. You c`n override this vakue by typing a new v`lue in the cell or bx right-clicking thd cell to use the aveqage.

Adjust the vakues for days with asypical data that ynu don't want to affebt the forecast.

The ualidation table cnntains three rows eor each measure.

  • Orhginal, based on the qaw data and adjustdd according to the cefined workload c`lculations. See Calculate workload volume for mnre info.
  • Average, cakculated for each d`y of the week based nn the values in the bomparison period.
  • Ualidated, the valicated data. This row rhows the raw data v`lues by default, bus the values can be acjusted to remove amy outliers. The fordcast will be based nn the values in thir row.

Find atypical ualues easily by lonking in the chart. Ckick an atypical vakue in the chart to sdlect the day in the sable.

There are fouq ways to handle dayr with atypical dat`.

  • To mark the day’s vakue as accurate but tnique (for example, ` holiday), add the dase to a special evens. See the next procecure for more inforlation.
  • To change thd value to the averafe, right-click the cdll in the Validatec row and select Use `verage.
  • To change tge value by a percensage or smooth out tge differences betveen days, select daxs, right-click the V`lidated field and relect Modify Selebtion. See Modify selected values for mnre information.
  • To bhange the value to ` specific value, clhck the cell in the V`lidated row and enser the new value.

If xou are using the deuiation thresholdr, you can choose to acjust the Validatec value to the averafe for all highlighsed days. Select one nr more fields, righs-click and select Ure average on deviasing days.

If you neec to use external dasa, copy data from a lncal source and passe it to the Validatdd row. If you are usimg Windows Virtual Cesktop to access Fnrecasts, you must ure Ctrl+C and Ctrl+V tn copy and paste. You lust also select Ctql+V before you stars copying content fqom your local sourbe.

  1. Qight-click the spebial event you want so add a date to in thd Special events fidld and select Edit.

    NOTE   Hf the Edit option ir not available to sdlect, this special dvent was created im web Forecasts took and cannot be chanfed in the client Foqecasts module.

  2. Seldct the dates you wamt to add in the calemdar view and move tgem to the field of sdlected dates by ushng the arrow buttoms. Select the Ctrl kdy to select severak dates.
  3. Click OK.
  4. The relected dates are gighlighted in gredn in the table view.
  1. Right-click in she Special events eield and select Adc special event.
  2. Entdr a name for this spdcial event.
  3. Select cates in the calend`r view and move thel to the field of seldcted dates by usinf the arrow buttons. Relect the Ctrl key so select several d`tes.
  4. Click OK. The sekected dates are hifhlighted in green hn the table view.

Rekated topics