Keplar.in Information

Keplar team uses Multinomial Logit, Poisson, complex Poisson, and some non-conventional models like Markov chain are also used to model consumer behavior. Our product range spans as follows: 1) Choice Modeling (Basic/general attraction models+NLM+RCSM+Markov Chain Model, threshold utility) 2) Assortment Optimization 3) Multiple Fare classes in single resource revenue optimization with heuristic protection levels 4) Sales based dynamic programming (based on Hamilton Jacobi Bellman Equation aka HJB) 5) Dynamic Pricing 6) Competitive Assortment and Dynamic pricing under competition Also looking forward to implementing some sort of location covering sets over the above class of products We focus on the following technology to model above: 1) Recurring Neural Networks (Also tried convolutional neural network to disappointment!!) 2) Lasso Regression Regularizations 3) Support Vector Machines 4) Clustering 5) And Ensemble bagging 6) Also looking forward to using Probabilistic Neural networks