#Video game recommendations tv#

#Video game recommendations tv#

The kind that’s referenced in TV shows and played even by non-gamers. Based on domain knowledge, the recommendations appear sensible. Monument Valley and especially its sequel, Monument Valley 2 is one of those rare video games that has crossed over into the mainstream.

We would need to conduct A/B testing to see if it outperforms the existing similar games recommender. Back when games were primarily sold on shop floors, and the digital age of gaming was a farfetched pipe dream, there was no such thing as an indie game. Once integrated with the Steam interface, this can be used directly to display similar games on a given game's page. We looked at item similarity and built an 'item-to-item' recommendation enginem, which generates k similar items using cosine similarity. Based on domain knowledge, the recommendations appear sensible. We would need to conduct A/B testing to see if it outperforms the existing games recommender. Once integrated with the Steam interface, this can be used directly to display recommended games on a user's homepage. The Angry Birds franchise has expanded far and wide, to the point where its hard to find a game now that doesnt appeal to mobile gamers. We built a user game recommendation engine, which takes in a user id and generates k games that the user is predicted to like. We would recommend ensuring advertisement deals are priced at a premium during this period. We saw that October, November and December have the highest number of game releases. Other recommendations can be found in the Notebooks.

We drew insights and provided recommendations to the marketing team, such as the below regarding timing of game releases. It is aimed to be between 5 and 10 minutes long.Ī collection of helper functions were defined in the project and saved in resources.py for future use.Ī blog post on Medium was created as part of this project. If you have any recommendations I am open to them. The presentation gives a high-level overview of our approach, findings and recommendations for non-technical stakeholders. To help in purchasing the game, I have either added an Amazon, PSN or Steam link to direct you to the store. The Jupyter Notebook 0_Master_Notebook should be viewed first and contains links to the other notebooks. Jupyter Notebooks are our key deliverables and contain preprocessing, exploration and modelling. Steam (hypothetically) currently uses a content based model to genereate recommendations and is keen to see which method performs better.įinally, I will be looking at item-to-item recommendations, providing similar items which can be listed on a game's page. By Nadine Amersi-Belton Problem statementĪs a (hypothetical) junior data scientist at Steam, my role involves drawing insights from data and creating data-driven tools to boost sales and revenue.įirstly, I have been tasked with providing useful recommendations for the marketing team, who are currently reviewing their avertising contracts with various publishers.Īs a second component, I have been asked to build a user recommendation engine using collaborative filtering methods to incentivise purchases.