Our demand forecasting and predictive modeling services allow clients to rapidly assess key drivers, their incremental impact on a measure of interest. These methodologies and frameworks are customized to an individual client context (no “one size fits all”) and industry. We work jointly and collaboratively with the client team to collect potential data for model development and forecasting. We adhere to all commonly accepted practices for model training and development. Through our academic and research tie-ups we are able to apply multiple frameworks and methodologies. For clients without their own analytics infrastructure, we can help develop these toolkits using our infrastructure allowing the client to develop a proof of concept with lower investment costs and then scale up in a phased manner. To ease implementation, we can also provide dashboards and scenario planning tools to understand the impact of different levers.

Our Demand Forecasting and Predictive Modeling capabilities include:

• Clustering techniques including k-means, Hierarchical clustering
• Machine learning – Supervised as well as unsupervised – including artificial neural networks, support vector machines, gradient boosting
• Discrete choice model, multinomial logistic-nominal, ordinal model development
• Finite mixture and mixed models
• Instance based algorithms including k-nearest neighbor, self-organizing maps
• Regularization Algorithms including ridge regression, Least Absolute Shrinkage and Selection Operator (LASSO), least angle regression (LARS)
• Naïve Bayes, Multinomial naïve Bayes, Bayesian Network
• Dimensionality reduction algorithms including Factor Analyses, Principal Component Analyses, Linear Discriminant Analyses
• Ensemble Methods combining different techniques