USING LEARNED APPLICATION FLOW TO PREDICT OUTCOMES AND IDENTIFY TROUBLE SPOTS IN NETWORK BUSINESS TRANSACTIONS

An approach is provided that receives, over a computer network, transaction data from a number of clients that are running an app. The approach generates association rules by inputting the transaction data to an association rule learning algorithm, such as an Apriori algorithm. Each association rule is based on a user transaction pattern and a desired result, and each association rule includes a generated confidence value that pertains to an expected performance of one of the steps included in the respective association rule. The app is then modified based on an analysis of the generated confidence values, with the app modification being directed towards improving one or more of the confidence values.