As part of your data cleansing steps, you might need to remove duplicate rows of data from your dataset.
Validate Duplicate Data
In some cases, it might be acceptable to have duplicated data. For example, additional records using the same primary key might be included in a dataset as amendments or detail records.
NOTE: Before you remove duplicates from your dataset, you should verify that the data should not contain duplicates at all. If the data structure supports some duplicate elements including key values, you should exercise care in how you identify what constitutes duplicate information.
Remove duplicate rows transformation
|D s product|
Tip: If you are attempting to identify if there are duplicate rows, check the row count in your dataset before and after you have added this transformtransformation.
- This transform transformation is case-sensitive. So, if a column has values values
HELLO, the rows containing those values are not considered duplicates and cannot be removed with this transformtransformation.
- Whitespace and the beginning and ending of values is not ignored.
Before applying the the
Remove deduplicate transform
rows transformation, you should attempt to normalize your data. You can use the following techniques to normalize a few columns of data.
NOTE: If you have more than 20 columns of data, you might be better served by trying to identify a primary key method for de-duplicating your dataset. Details are below.
For individual columns, you can use the the
trim function to remove leading and trailing whitespace:
NOTE: To preserve the original column values, use the
Since the the
Remove deduplicate transform
rows transformation is case-sensitive, you can use the the
LOWER function function to make the case of each entry in a column to be consistent:
Deduplicate Rows Based on a Primary Key
An easier method to Another method to deduplicate data might be to delete rows based on one or more columns that you identify as a primary key for the dataset. A A primary key is key is an identifier that uniquely identifies a row of data within a dataset. It can be a single field (column) or a combination of columns. For example, in a datasets of restaurant locations, the primary key can be a combination of RestaurantName, Address, and Zip.
NOTE: Before continuing, you must identify a primary key for your dataset. See See Generate Primary Keys.
When you have identified your primary key, you should identify the appropriate method for your dataset. Please complete the following steps.
If your primary key spans multiple columns, use the
Merge columnstransformation to bring the values into a single column:
D trans Type step p01Name Columns p01Value RestaurantName,Address,Zip p02Name Separator p02Value '-' SearchTerm Merge columns
Rename the generated column:
Use the following transform transformation to generate a new column, comparing each value in the the
PrimaryKeycolumn column to the previous one:
D trans Type step p01Name Formulas p01Value PREV(PrimaryKey, 1) p02Name Order by p02Value PrimaryKey SearchTerm Window
For each row, the value of the new column is the value in the the
PrimaryKeyfor for the previous row. Now, test if this value is the same as the value in the the
PrimaryKeycolumn column for the current row:
D trans p03Value IsDupe Type step p01Name Formula type p01Value Single row formula p02Name Formula p02Value IF((window==PrimaryKey),true,false) p03Name New column name SearchTerm New formula
The new column (
IsDupe) contains contains
truefor for duplicate primary keys. Delete the rows that are duplicates:
D trans Type step Description Condition p01Name (IsDupe==true) SearchTerm Delete rows
- Drop Delete any generated columns that are no longer needed.
While this form of duplicate data is rarer, you might want to check on the possibility of duplicate data between your columns. To check for duplicate column data, you can use a transform transformation similar to the following:
In the generated column, values that are are
true indicate duplicate data. If all values are are
true, then you can remove one of the columns.
|D s also|