Massively by HTML5 UP

Excel

Exploratory analysis of GameCo’s video game dataset

It is October 2016 and GameCo’s executive board is planning the marketing budget for 2017. They’re assuming that sales for the various geographic regions have stayed the same over time, and they’ve asked the data analyst to look into the data to see if this is still true. If it's not, the marketing budget will need to be redistributed among the regions to maximize return on investment.



Objective:

Descriptive analysis of a video game data set to foster a better understanding of how GameCo’s new games might fare in the market.

Key Business Questions:

  • Differential game popularity
  • Main competitors per markets
  • De/increased games‘ popularity
  • Varying sales figures between geographic regions over time

The dataset

Open data set, retrieved from

VGchartz Dataset

Covers historical sales of video games.
Tracks the total number of units of games sold (not financial figures) from 1980 to 2016.

Procedures:

  • Excel:
  • Data preparation
  • Grouping, summarizing data
  • Descriptive analysis
  • Visualizing
  • Stakeholder Presentation in PowerPoint

Analyses & Results:

Results & Recommendations:

Other than expected, the EU market was the most important in 2016. It eclipsed the NA market by approximately 6% of units sold in the global market. The regional market differ in most important genres:
  • NA: Fighting, Shooter, Platform
  • EU: Racing Strategy
  • JP: Simulation, Miscellaneous

The Japanese market holds 13% of shares in the global market, but it behaves differently from the other markets and the global market: More research necessary.

Highly diversified, fluctuating markets

  • Concentrate on the EU market and here, Racing and Strategy Games in budget planning for 2017
  • Find out what makes the Japanese market special, invest in its particularities
  • In general, dive deeper into the specificities of each market and diversify the marketing strategies accordingly
  • Investment in deeper data analyses; for future data collections, more demographics of customers and data on customer behavior can help to observe and forecast regional and global trends

Find the Stakeholder Presentation here:

Stakeholder Presentation

Future approach/ improvements with Python-coded Machine Learning algorithms:

Time-series analysis: Predicting future sales (next month, next year) based on past trends; applying moving average.

Classification: Place yearly sales value in bucket; i.e. Will yearly sales value be above (class 1) or below (class2) a set threshold/target value?

Regression: Predict raw future sales values using various regression techniques—line of best fit.