Mobile Gaming Usage Intelligence Modelling

Project Description

We are on a mission to build the next generation of mobile game usage intelligence to help to identify the most valuable users.

Real iOS Gamer Activity Intelligence, Not Just Installs & Ranks 
Have your ever wanted to compare mobiles games usage and base your decisions on how players really play and not only on how games are ranking in the app stores?
With no SDK, Whally tracks anonymously 400+ million players posting daily more than
80+ million scores within 20 game categories on 190,000+ iOS mobile games.

Whally.com has released 2KPI's to date : 
- Whally Index show a iOS game index based on daily player activity for games that offers public leaderboards
- Whally Overlap shows where players are across mobile games, analysed daily

• The problem that you are trying to solve 

Whally is tracking a subset of best iOS games around 25% of all 150000 best games. All data collected should provide us with enough material to extrapolate results for all 150 000 games. The goal is to provide an estimate of active scorers for every game in AppStore.

• The kind of expertise you require 

Whally needs expertise in data development along with notions in mathematics to build a interpolation and extrapolation tool to estimate scorers activity for all iOS games. The tool needs to take into consideration app ranks, number of active players and computes an estimate for missing data points along with the error margin associated.

• The data sources at your disposal and their formats (Pricing data in CSV format, etc.) 

All the data is crawled from the sources into a Cassandra database. A few tables cover the data sources, including the list of applications with their associated ranks per day (segmented by game sub-category (trivia, action, sport, etc), model (free, paid, grossing) and countries), and also the associated active players per application and per day.

• What is your current technology stack? (D3.js, Ruby on Rails, Hadoop, etc.) 

Data storage : Cassandra cluster
Data processing : Apache Spark (jobs currently written in scala, but python jobs are ok), Pandas for python/scikit-learn

• What is the deliverable? (Algorithm, dashboard, web application, advisory service, etc.) 

Deliverable will be Apache Spark jobs in Scala/python or Pandas/scikit-learn Python code that Whally can deploy to ensure daily processing.

• Does the deliverable need to be deployed in the cloud or your infrastructure? 

No, Whally will cover this part.

• Attach a sample data file, if you can 

Not really applicable, but schema of the database can be provided, though.