With 250+ Kiosks across the United States, the client offers on-the-go mobile charging devices on rent or buy options. A customer can visit any kiosk, choose rent or buy, select device type, and swipe card to pay. When the customer is done with the charging, he needs to return the rental charger to any nearby kiosk. All Kiosks’ transaction data is stored in a semi-structured database in MongoDB. The client wanted to draw a meaningful insight out of this huge volume of data to maximize ROI.

Segmentation Model
  • Looked at a total of 132 variables to be narrowed down to 31 to build a segmentation model
Clustering
  • Used a k-means algorithm which considered all of the different attributes to group the most similar kiosks together, split across 12 clusters
Recommendations
  • Set out the data challenges faced by the client and provided recommendations to improve the efficiency and accuracy of analysis

Solution outcomes

Predictive modeling
  • Revenue trend forecast
  • Market basket analysis
  • Predictive maintenance and up-gradation by kiosk and by location
Daily, Weekly, Monthly, Quarterly and Yearly Average revenue and transactions per kiosk
Actual vs. Budget revenue and transaction analysis
Inventory analysis that shows the movement of inventory throughout the kiosk network
Drawing the various patterns of kiosk networks’ usage for efficient servicing, and process improvement
Repeat Customers analysis
Kiosks KPI reporting