Customer Size: Large organization
Domain: Event Management
Technology: Python, Flask, Jinja2
Plugins: REST API, JSON
Web Service: Service Integration with Web portal and Mobile app
Algorithm: Decision Tree [Modeling of algorithm using Scikit Learn]
A US based leading analytics company sells point-of-sales solutions that is easy to install and simple to use and providing immediate insights for the businesses and generates maximum revenue.
Our Client wanted to develop a inventory based POS solution for caterers and Event Agencies to forecast food consumption at any kind of event and minimize food wastage, the solution was suppose to very much flexible with various dependencies such as weather, Nature of event, level of event organized etc…
Our client primarily asked us to make a solution that could achieve given below business needs;
- Real time food predictions,
- A highly interactive dashboard to deliver insights,
- Easy Integration with Web portal and Mobile Apps.
Developing solution that forecast is never easy and routine task, so we decided to send our best team at client premises to collect insight and grip over the standard process, we also looked into historical data and business model to create a robust point-of-sales solution. The features taken into consideration during the algorithm formation were number of tickets sold in past, weather, and type of event, stadium/ wing capacity and total tickets sold till date
Performing data assessment, data preparation, data modelling, algorithm modelling and integration of score model on the accumulated data created linear Regression and Decision Tree Algorithms. Through these algorithms, a predictive solution came in to picture, which was capable to forecast future orders at the event, crystal reporting, and customer behaviour analysis to evaluate business rhythm
Our System delivered a high end insight data to support client DSS process, improved efficiency on operations along with drastic efficiency over supply and chain management. At present, the solution is implemented at a much wider network of food counters during various events, and the forecasting it delivers is accurate by 96.45% (Based on clients input).