LodgIQ uses multi-textured data feeds to create a proxy for demand. Data is ingested, wrangled and aggregated on a Big Data stack in a cloud environment. Machine Learning models are run on a daily basis to make accurate market-level and property-level forecasts and provide pricing recommendations. This is done for 100+ global markets and several hundred properties.

Models are trained and tuned on a continuous basis. Most of these model building and tuning tasks are extremely computationally intensive. Care is taken to bid for cloud compute instances, acquire, setup and run the models on cloud based compute instances and relinquish them as soon as the jobs are finished. All of these are accomplished in a programmatic fashion. This is necessary to run a spectrum of state of the art models from Tree Ensemble to Deep Learning models, accomplishing a high level of accuracy at a very affordable cost.