I am sharing a data product http://ohrivikas.weebly.com/forecasting.html(select Market basket analysis option) for doing Shopping basket analysis.
This application suggests items for cross selling when a user selects any item from the buy or shopping items as part of order, as shown in following image.
![Image1.PNG]()
By default it is using Groceries data set containing 30 days of real-world point-of-sale transaction data from a typical local grocery outlet with the items that are aggregated to 169 categories.
Item suggested is based on likely hood of possible based on what product is currently in the order and previous history of orders.
Choosing Item associations tab, one can view all the possible item suggestions learned from historical sales transactions, indicating strength of relation between buying of certain items leading to purchase of subsequent items. These can be analysed & filtered narrowed by appropriate confidence, support and lift factors as shown in following image.
![Image2.PNG]()
It can accept user input data two column file in XLSX format as well as new data to create item association rules for shopping basket analysis and cross selling suggestions. For a SAP system it can be easily created by downloading Sales Items from table VBAP using Order Number as unique Transaction ID and Material as the Transaction Item two columns as shown in the image below.
| Order | Product |
| 5728 | R-1001 |
| 5728 | R-1111 |
| 6727 | R-1120 |
| 6730 | R-1130 |
| 6730 | R-1140 |
This endeavor to build this appication started quite a while ago when i developed HANA based application available at https://s1hanaxs.hanatrial.ondemand.com/p202072trial/ohri/PAL_01_settings/ to do Market or shopping basket analysis, using HANA Predictive Analysis library Apriori functions & SAPUI5 framework. Some of the background information on this application including explanation of terms confidence, support & lift and shopping basket analysis is shared in my earlier earlier post Cross Selling and Shopping Basket Analysis - HANA Cloud Platform Predictive Application.
It's logical to bundle this application as data product for end user . Since SAP HANA XS Administrator access for free trial version is unavailable to setup SQLCC, connection configuration so I had too ask for user's SCN or SAP ID and used it to add to list of authenticated users who could use this application and for any other user of the application, it errors out for lack of valid authorizations.
Also sometime earlier i's able to connect to HANA cloud trial version using JDBC connection as shared in my post R Integration with HANA with JDBC - Test of "Outside In Approach" - Part 2 giving me idea to build data product using a different interactive graphics framework using javascript libraries like HighCharts, morris, nvd3, polycharts ,dimple & even open map leaflet libraries. Incidentally SAPUI5 is built over some of these same libraries. Further this approach opens up more possibilities for data (including big data) analysis and machine learning with ERP Business data. Little digression of thoughts ,In my opinion big data term is often understood with data of large or high volume scale , though true as it warrants then distributed & parallel computing platforms, but big data applications may include "small" data as well , term big is relative , data that can not be easily comprehensively dealt with in given computing resources can be regarded big either due to its underlying structure and or speed at which it changes.
Which are most frequent or least selling items can be analysed interactively by selecting Frequently Sold Items tab as shown in image below.
![Image3.PNG]()
Data product for shopping basket analysis is now easily available for end user by clicking Market Basket option at http://ohrivikas.weebly.com/forecasting.html
