Validate historical data

Validate the historical data from the ACD platform to create a reliable base for forecasting. In the validation you work through the historical data in order to identify data that is not representative. This is essential to ensure the quality of the forecast.

During the validation you detect days that deviate from the normal workload. This can be days that are abnormal, such as a snowstorm or a technical issue causing an increase in traffic. These values can be corrected manually.

Other days that deviate from the normal workload are recurring, such as national holidays. These can then be saved as special events and reused later. If an outlier in the data is because of a recurring event, for example a public holiday, it can be added as a special event. Special events have two purposes; to exclude the day from the normal set of data to base forecasts on and to collect a special set of data to forecast for future dates of these recurring events.

The values in the Validated rows are used to create the forecast and to calculate seasonal variations and trends.

IMPORTANT   Each period only needs to be validated once. Ensure to validate historical data regularly. Create a process to for example in the beginning of each month go through the data of the past month. Over time you collect a large amount of validated data to base your forecasts on.

Prerequisites

Page location

Client > Forecasts > Prepare workload > Validation tab

Procedures

  1. Expand the skill you want to work with to show the workloads.
  2. Right-click the workload to validate data for and select Prepare workload.
  3. Continue to select a period to validate.

Select the period to validate. When working with a shorter period, for example one month, you can use the same period for validation and comparison. The averages are then calculated for the period that you validate.

  1. Select a start date and an end date in the Validation period fields to define the period to validate data for.

  2. Select a start date and an end date in the Compare with fields to define a comparison period.

    The data in the comparison period is used to calculate an average for each day of the week. You can compare the data in the validation period to these averages to help you find outliers. For the comparison period to be useful, it must be similar to the validation period regarding for example volumes and average talk time. Do not use a too long period as the comparison period, as those averages won't be significant to the data you are validating.

    If you use the averages, continue to define deviation thresholds. If you don’t need to compare the data in the validation period to averages to find outliers, you don't need to define a comparison period.

  3. Click Apply to load the data for the selected periods.

The deviation thresholds help you spot outliers in the data. Original values in the table that deviate more from the average than the defined percentage are highlighted in red. The names of the fields depend on the type of skill. This procedure is based on a telephony skill, but the functionality works in the same way for all types of skills.

  1. Enter the deviation percentage in the Deviation calls field.
  2. Enter the deviation percentage in the Deviation talk time field.
  3. Enter the deviation percentage in the Deviation ACW field.

Adjust the values for days with atypical data that you don't want to affect the forecast.

The validation table contains three rows for each measure.

  • Original, based on the raw data and adjusted according to the defined workload calculations. See Calculate workload volume for more info.
  • Average, calculated for each day of the week based on the values in the comparison period.
  • Validated, the validated data. This row shows the raw data values by default, but the values can be adjusted to remove any outliers. The forecast will be based on the values in this row.

Find atypical values easily by looking in the chart. Click an atypical value in the chart to select the day in the table.

There are four ways to handle days with atypical data.

  • To mark the day’s value as accurate but unique (for example, a holiday), add the date to a special event. See the next procedure for more information.
  • To change the value to the average, right-click the cell in the Validated row and select Use average.
  • To change the value by a percentage or smooth out the differences between days, select days, right-click the Validated field and select Modify Selection. See Modify selected values for more information.
  • To change the value to a specific value, click the cell in the Validated row and enter the new value.

If you are using the deviation thresholds, you can choose to adjust the Validated value to the average for all highlighted days. Select one or more fields, right-click and select Use average on deviating days.

If you need to use external data, copy data from a local source and paste it to the Validated row. If you are using Windows Virtual Desktop to access Forecasts, you must use Ctrl+C and Ctrl+V to copy and paste. You must also select Ctrl+V before you start copying content from your local source.

  1. Right-click the special event you want to add a date to in the Special events field and select Edit.

    NOTE   If the Edit option is not available to select, this special event was created in web Forecasts tool and cannot be changed in the client Forecasts module.

  2. Select the dates you want to add in the calendar view and move them to the field of selected dates by using the arrow buttons. Select the Ctrl key to select several dates.
  3. Click OK.
  4. The selected dates are highlighted in green in the table view.
  1. Right-click in the Special events field and select Add special event.
  2. Enter a name for this special event.
  3. Select dates in the calendar view and move them to the field of selected dates by using the arrow buttons. Select the Ctrl key to select several dates.
  4. Click OK. The selected dates are highlighted in green in the table view.

Related topics