Trend and Share Trend grids show data values for the individual dates in a range of days, weeks, months, or periods.
In addition, a row or column may show daily, weekly, monthly (or period) averages across the whole date range. These averages use a "simple average" calculation that divides the total for the entire date range by the number of days, weeks, months, or periods (depending on the selected resolution) in the date range (for example, total value/16 weeks).
Another row or column may show the totals, which are calculated in one of the following ways to provide the most meaningful results depending on the type of data.
Total for entire time period (additive): This type shows the total value for the entire date range. These totals are used for data that adds up meaningfully over time, including volume, revenue, cost, margin, and frequency measures that don't use multiplication or division.
Calculated from component total values: Many measures don't add up meaningfully over time (for example, per unit revenue or costs). Frequency, revenue, cost, and margin measures that use multiplication or division in their formulas fall into this category. For these measures, the total calculation first finds the total for each measure in the formula and then performs the calculation using the total values. For example, depending on your organization's definition, UXT might calculate the average Per Unit Cost by dividing the total cost for the whole date range by the total number of units for the whole date range. In this way, UXT weights the results by the data, so the time periods with more data have a greater effect on the end results.
Non-transactional data (not unique counts): Non-transactional data doesn't add up meaningfully over time, because it represents a condition that spans a number of days, weeks, or months. Therefore, the values are not simply summed across the date range to reach the total. Instead, the total row or column shows the average results for the whole date range, weighting the data by the number of days a condition spans. For example, a month that has more days would have a greater effect on the end results. Non-transactional data includes data or record counts from a non-transactional data cube, which belongs to the causal data category. However, this classification of data does not include key count measures, which are discussed below.
Unique counts: Your organization might use a measure that counts the unique occurrences of members in transactions. For unique counts, UXT only counts each member once, so the values don't add up meaningfully over time. Instead, the total row or column shows the cumulative results for the entire date range, counting each member only once. For example, if member ABC was in a transaction during the first and last week of a six-week date range, UXT would count ABC once. Also, this type of data will not be available in range windows that show a subset of the date range. This type of data belongs to the causal data category.