Reacting to Variations in Product Demand: An Application for Conversion Rate (CR) Prediction in Sponsored Search

In online internet advertising, machine learning models are widely used to\ncompute the likelihood of a user engaging with product related advertisements.\nHowever, the performance of traditional machine learning models is often\nimpacted due to variations in user and advertiser behavior. For example, search\nengine traffic for florists usually tends to peak around Valentine's day,\nMother's day, etc. To overcome, this challenge, in this manuscript we propose\nthree models which are able to incorporate the effects arising due to\nvariations in product demand. The proposed models are a combination of product\ndemand features, specialized data sampling methodologies and ensemble\ntechniques. We demonstrate the performance of our proposed models on datasets\nobtained from a real-world setting. Our results show that the proposed models\nmore accurately predict the outcome of users interactions with product related\nadvertisements while simultaneously being robust to fluctuations in user and\nadvertiser behaviors.\n