ACRE analysis is done using pipeline datasets. These contain Input Columns of user data and Pipeline Columns, which are calculated from previous columns by Pipeline Tasks. Each Pipeline Task performs a particular type of analysis (count, extract, tag by rules, tag by machine learning, etc.) on a previous column, using a stored set of parameters and generates one or more columns of results.
For example, in the pipeline dataset shown above, the 4 blue Input Columns are loaded from an existing spreadsheet of lab data. ACRE then calculates the contents of the 6 green pipeline columns by executing 4 pipeline tasks. Each pipeline task has:
· A Task Type which can be: Count, Extract, Edit, Lookup, Tag by Rules, Tag by Machine Learning.
· A Task Name
· Depending on the Task Type, a set of results columns, each of which has a Column Type and Column Name.
An unlimited number of pipeline tasks can be executed. Tasks can be added/deleted/modified at any time and all dataset columns will be updated. Additional input data can be loaded to the dataset at any time and the corresponding pipeline columns will be automatically created. Dataset contents can be re-calculated via Update Jobs.