Product Recommendations

This data set contains retail transactions in the receipt format. While more difficult to manipulate, the receipt format is particularly helpful for sales data because it allows you to attach multiple items from the same category to each transaction through the transaction_id. This feature makes the receipt format ideal for basket analysis.

The sample data set contains five columns: the transaction_id, item_id, transaction_date, item_name, and item_category. Only the transaction_id and item_id are required fields, but the added transaction_date, item_name, and item_category give added information to be included in the analysis.

Note that the receipt format does not permit user-defined column headings. Some retail data sets include additional columns that relate more to the product than to the purchase, like product margin. These columns can be moved or rolled into the receipt format using the transaction_id. 

If your data looks similar to the Receipt Data format, you may only have to rename the column headers or change the column order. Other data will require more work. If you need assistance with the format, first refer to http://emcien.helpgizmo.com/help/article/link/preparing-your-data

You can upload data files to Emcien using SFTP and your preferred FTP client, such as FileZilla or Cyberduck.

To connect to Emcien using your FTP client, use the following credentials:

• Host: feeds.emcien.com
• Username: {Your Emcien Feeds server Username}
• Password: {Your Emcien Feeds server Password}

1. Using your preferred Internet Browser, navigate to the Emcien Sign In page:

   • For production users, navigate to http://patterns.emcien.com/

2. On the Home page click Retail Sales.
   
   

3. If you believe that your data may be incomplete, select the button Override Defaults and click Allow Limited Invalid Lines. This will increase the fault tolerance of your analysis by omitting as many as 10% of the uploaded lines of data.

4. Click on the green “Analyze Data” button to begin your analysis. 

The Dashboard page for the Retail Sales data set is displayed. On the Dashboard page you can view the number of connected data values, the number of transactions analyzed, and more.

1. On the Dashboard page click Clusters.

 

The Clusters page is displayed showing an overview of the connections in the data set. 

2. On the Clusters page click A-Z to sort the clusters alphabetically.

3. Click the Adhesives category.

The categories connected to the Adhesives category is displayed. In this data set, adhesives are connected to other adhesives, tape, and pens.



You can also see that in the Adhesives category glue guns and the sticks for the glue gun are connected.

4. On the Clusters page, click on Colored Pencils & Markers.

Items in the Colored Pencils & Markers category connected to items in the Adhesives category are displayed. You can use the connections on this page to find point-to-point connections.

 

For example, transactions containing themebook purchases also frequently contain crayon, marker, filler paper, and glue purchases. You can use this information for cross-selling or to put together product bundles. 

5. Click the GLUE;ELMERS;4 OZ;WE item.

The Item Detail page for the GLUE;ELMERS;4 OZ;WE item is displayed. On the Item Detail page you can see transaction trends for the GLUE;ELMERS;4 OZ;WE item over time.



You can also view items frequently connected with the GLUE;ELMERS;4 OZ;WE item.

6. Click Explore Graph in the right sidebar.

A graph containing connections to the GLUE;ELMERS;4 OZ;WE item is displayed. Note that all of the purchased products are represented in this analysis, allowing these outputs to drive recommendations in real time. 

7. Hover over the element on the graph marked THEME BK WR 1-SUB 92 BRIGHT item.

Items connected to the THEME BK WR 1-SUB 92 BRIGHT are highlighted.



Each of these product relationships can be analyzed automatically to deliver intelligent recommendations to customers in real time through an Emcien API. For more information on Emcien APIs click here.