Cleansing data

Use this procedure to clean up data before processing to ensure that you receive the most helpful recommendations.
  1. Sign in as Application Administrator or Supply Management Administrator.
  2. For Application Administrator, select Supply Management > Inventory Intelligence > Configuration.
    For Supply Management Administrator, select Inventory Intelligence > Configuration.
  3. Select the Data Cleansing tab.
  4. In the Analysis Time Window section, specify this information:
    Start
    The engine recommendation is based on an analysis during a specific window of time. The Start parameter indicates how far back the analysis window starts. We recommend 364 days so that the analysis window is a history of transactions that occurred in the previous year.
    End
    Select a few days in the past for the end of the analysis window to ensure that transactions are complete.

    For example, you schedule recommendations on the fourth of every month. We recommend that you select three days before the latest transaction. This practice is to ensure that the entire previous period is included.

  5. In the Daily Data Filtering section, specify this information:
    Omit transactions prior to last stock UOM change
    When the stocked item unit of measure changes, the transactions that occurred before that change are not compatible with values that are recorded after the change. These older transactions can almost always be omitted. Select this check box to omit them.

    Daily data pertains to transactions that have occurred during a specific day or status as of that day. Because data quality is always an issue, you can set up thresholds to find outliers. When an outlier is encountered, the transaction is excluded from analysis.

  6. In the Quantity Cutoff section, specify this information:
    Issues On-hand
    Specify a number to represent the high level threshold.
    Issues
    Specify a number to represent the high level threshold.
    Receipts On-hand
    Specify a number to represent the high level threshold.
    Receipts
    Specify a number to represent the high level threshold.

    Real-world data can contain errors such as typos. If a typo exists where the daily data value exceeds the threshold, then the transaction is omitted.