Exploratory analysis · 2025
Video Game Sales
Which platform and genre should new investment back, and does the answer change by region?
Python · Pandas · Matplotlib · SciPy
The question
Which platform, which genre, and — separately — which regional strategy make sense for new game investment, based on the sales record rather than intuition.
The data
16,000+ titles (16,715 rows) spanning 1980–2016, with per-region sales (NA, EU, JP, other), critic and user scores, genre, platform and ESRB rating.
Method, and why this method
Column names were standardised to lowercase; year_of_release was cast to Int64 and
user_score from object to float (pd.to_numeric with errors='coerce', which turns the
"tbd" placeholder into NaN so the column behaves numerically). total_sales was derived as
the row-wise sum of the four regional sales columns, to enable direct global comparison
between titles. Missing name/genre values (2 each) were filled as a placeholder; missing
critic_score, user_score and rating (8,578 / 9,125 / 6,766 rows respectively) were left as
NaN and excluded per-analysis rather than dropped outright, since they most likely reflect
insufficient reviews rather than bad data.
Because the goal is to inform a forward-looking investment decision rather than describe gaming history, the platform, genre and regional analysis was restricted to 2013 onward (2,233 of the 16,715 records), ensuring greater relevance and adherence to the current market.
Two hypothesis tests closed the analysis, both two-sample t-tests run with equal_var=False
after checking that the compared groups' variances differed:
- Xbox One vs PC average user rating (variance 1.91 vs 3.04) — p = 0.1476 against alpha 0.05: H0 not rejected, no evidence the average ratings differ between the two platforms.
- Action vs Sports average user rating (variance 1.77 vs 3.18) — p ≈ 0.0000 against alpha 0.05: H0 rejected, users rate the two genres differently.
Findings
Across the full history, releases climbed steadily from the 1990s on, peaking in 2008 (1,427 titles that year) rather than in the most recent years — a possible shift toward fewer, higher-quality releases.
From 2013 onward, PS4 leads total sales at 314.14M, ahead of PS3 (181.43M), Xbox One (159.32M), 3DS (143.25M) and Xbox 360 (136.80M); PS4 also leads on average sales per title (0.80M). On PS4 specifically, critic scores correlate with sales more than user scores do (critic score × sales = 0.41, user score × sales = −0.03), but neither correlation is close to 1.0 — other factors such as marketing, franchise, exclusivity and price likely carry more weight than the scores themselves.
By average sales per title in the 2013+ window, Shooter leads (~1.25M per title, credited to franchises like Call of Duty), followed by Sports and Platform. Action leads in both title volume and total sales but ranks only 7th by average per title — a large, fragmented, highly competitive genre rather than a per-title winner.
Regional profiles diverge sharply. NA and EU both favor Action and Shooter, with the Mature (M) ESRB rating leading sales in both regions. Japan breaks the pattern: Role-Playing leads by a wide margin, and the Teen (T) rating leads sales there instead of Mature.
The decision it enables
For new investment: PS4 is the safest, most lucrative platform in the period studied. Shooter offers the best average return per title, while Action guarantees volume. The western market (NA and EU) rewards Mature-rated action and shooter titles, while Japan needs a separate strategy built around role-playing games.