Versions Compared
Version | Old Version 3 | New Version 4 |
---|---|---|
Changes made by | ||
Saved on |
Key
- This line was added.
- This line was removed.
- Formatting was changed.
Insert excerpt | ||||||||
---|---|---|---|---|---|---|---|---|
|
Introduction
Merges can be used to consolidate records from two or more tables. In this page we will work through a simple merge scenario to explain how merges work. All resource used in this example are available here:
Scenario
We have two sets of company information, Businesses 1.xlsx and Businesses 2.xlsx, that we want to merge together and remove duplicates. There are a number of challenges we must resolve to achieve this:
- The attribute names in each file are different.
There are duplicate companies that exist in both files.
There are duplicates within the Businesses 2.xlsx file.
Solution
Import the data
Using a file collector import each of the Businesses Excel documents into their own table and run analysis on each to populate them with data.
Make both of the tables static using the
option from the table hover over toolbar.Insert excerpt _static _static nopanel true This ensures we only read the files once.
Create the Merge Table
From the Create section of the analysis toolbar, click
Insert excerpt _datasource_show_tables _datasource_show_tables nopanel true drag the Merge Table onto your model.
In the settings, set the Name for example
Businesses Merged
The attributes can be created manually using the plus button in the Attributes section. Here we will use the attributes from the Businesses 1 table to automatically setup our merge attributes.
Hover over the Businesses 1 table and from the hover over toolbar click
.Insert excerpt _attributes_show_hide _attributes_show_hide nopanel true Tick all of the attributes and then drag them into the attributes section of the Merge Table.
- If PhixFlow asks for confirmation. Select Use original attribute name. Then click the arrow Image Added to confirm.
Image Added
Click OK to save and close the settings for the merge table.
Connect the Inputs
In the analysis model, hover over the Businesses 1
table
.From the hover over toolbar, select Connector, then click the end of the arrow onto the merge table. This connector will pipe data into the merge table.
A properties window opens for the pipe, set the name to B1 to better indicate the use of the pipe.
In the Sort/Group section of the properties click
and PhixFlow displays the list of attributes (columns).Insert excerpt _attributes_show _attributes_show nopanel true Drag the
Name
attribute into the Sort/Group section. This defines the attribute that will be used to merge on, we call this the key. Keys are listed in order and we reference them by number, therefore the keys must appear in the same order on all pipes leading into the merge table. This means attributes we are merging on do not need to have the same name.Similarly, add a connector from the table
Businesses 2
to the merge table and add the Company Name to the Sort/Group.Businesses 2 has duplicates within its data, we can define which records remove these and decide which record to keep. Here we will use the most recent
Onboarding Date
Onboarded Date to decide which record to keep.Drag the
Onboarding Onboarded Date
attribute into the Sort/Group section.Double click on the attribute and untick group Group and set Direction to
Z-A
. What we have done here is to use th By unticking group we are telling PhixFlow to sort by this attribute but omit it from grouping.PhixFlow will group and sort on the Company name , and use the
Onboarded Date
to also sort the groups. We need to select the top record from the group, and we do this by setting the Maximum Number of Records Per Group to 1. This will ensure we get the most recently onboarded business. The configuration should look like this:Image Added
Configure the Merge Table
Finally we We now need to tell the Merge Table what we are using to merge on, and which values from the tables we want to keep. For example, if there a company in both Business Tables, we can set the precedence to use the one from Business 1.how to populate its attributes.
- Click on the Merge Table, double click on the Name attribute. We are merging on the names set the first merge key we want to use, we only have one
Configure groupings on the pipes
- Double-click on the pipe from
CustomerAddresses
toCustomer All Details
to open its settings.- In the Basic Settings section, set:
- Name: this pipe is automatically named
in
.
- Name: this pipe is automatically named
In the Sort/Group section click Image Removed Show Attributes. PhixFlow displays the list of attributes (columns).
- Drag
CustomerRef
from the list into the settings. - Close the list of attributes.
- In the Basic Settings section, set:
- Double-click on the pipe from
CustomerAddresses
to
. This pipe will be calledSOURCE_CUSTOMER_PHONE_NUMBERS
in_2.
In the Sort/Group section, click Image Removed Show Attributes. PhixFlow displays a list of the attributes (columns) in the
CustomerAddresses
data.- Drag
CUSTOMER_REF
into the Sort/Group section. - Close the list of attributes.
- Now add attributes from
CustomerAddresses
to the merge stream:- Hover over
CustomerAddresses
and in the pop-up toolbar, click Image Removed Show Attributes. Drag all attributes from this list and drop them onto the stream icon for
Customer All Details
.How to select all the attributes in a list
To select all attributes, you can:
- either click the first attribute then Shift+click the last attribute
- or click the check box next to the Name header.
Image Removed
- PhixFlow asks for confirmation. If it is not already selected, select Use original attribute name. Then click the arrow Image Removed to confirm.
Image Removed
- Hover over
- Similarly, drag all attributes from
toSOURCE_CUSTOMER_PHONE_NUMBERS
Customer All Details.
- To see the resulting settings for
Customer All Details
, double-click its icon to open the settings tab and look at the Attributes section. You can see a list of 8 attributes (data columns).
Run analysis and view the stream data
- To run analysis on your new stream, hover your mouse pointer over the stream
Customer All Details
, and in the pop-up toolbar, click Image Removed Run Analysis. - To view the data you just loaded, hover over the stream and click Image Removed Show view. In the drop-down list, select Image Removed Default View. You will see the address data you loaded from the file combined with the phone numbers you loaded from the database
- the name of the business, enter
_key[1]
in the expression: - Image Added
- If you were merging on a second attribute enter _key[2], and this can be repeated for any number of attributes.
- For the Address1 attribute, we can decide on which source will have the best data to create a merged record from. Here we will let Businesses 1 take precedence. Double click on Address 1 and enter
ifNull(B1.Address1, B2.AddressLine1)
. This tell PhixFlow to use the Address from the Businesses 1, but if it is null then take it from Businesses 2. - Repeat these steps for the remaining attributes and your configuration will look like this:
- Image Added
- the name of the business, enter
- Run Analysis
- Run analysis to populate the Merge Table.
- Hover over the merge table and click
→ Default View.Insert excerpt _table_with_attributes _table_with_attributes nopanel true
Troubleshooting
- I need to clear the data? To remove data from a table see Removing Recordsets.
- I now have no data when I run my import? When you run an import the file you uploaded will be archived, therefore if you run the import a second time without uploading a new file there is nothing to import. Upload your file again and run analysis on the table.
- PhixFlow says it can't find the attribute. In the merge table ensure the values in your expressions match the names of the attributes coming from your input tables.