Okay, this might be an overkill method since we’re using machine + deep + reinforcement learning and might be complex to setup. If this is not what you want, feel free ignore it. Also, promotional content incoming!
Right now, what you want to do is to build a recommender system for your items, so that it only appears when the player most likely need that item. This should entice user without making it feel like that the player needs to buy this particular item in order to progress or win the game. Instead, it would “suggest” to the player that this item is best suited to improve the player’s situation.
With this recommender system in place, this should have its own UI, but not a popup that always block the player’s screen. You want it to be seamless as possible. So you put it in a shop or something.
Theoretically, you can use deep reinforcement learning models to build your recommender system, where it improves whenever the player buys the items. However, I expect extreme finetuning and training.
If you don’t want the finetuning and training part, you can use the old classic K-Nearest Neighbours (KNN) models, but you will fail to capture more complex decision making by the player.
DataPredict offers both of these models. In fact, it supports a wide range models for you to play around, so that it fits your development needs. With free commercial use for small game development studios, you can experiment freely without upfront costs. From simple KNN models to advanced reinforcement learning, DataPredict offers flexible tools to fit your development needs. Build smarter, player-friendly games with ease!