steam game data analysis
Dhruv Kapoor
2026-04-11
Steam data analysis
Dhruv Kapoor and Lakshay Mago
———————————————————
Load necessary Libraries
———————————————————
library(dplyr)##
## Attaching package: 'dplyr'## The following objects are masked from 'package:stats':
##
## filter, lag## The following objects are masked from 'package:base':
##
## intersect, setdiff, setequal, unionlibrary(ggplot2)## Warning: package 'ggplot2' was built under R version 4.5.3library(lubridate)## Warning: package 'lubridate' was built under R version 4.5.3##
## Attaching package: 'lubridate'## The following objects are masked from 'package:base':
##
## date, intersect, setdiff, unionlibrary(corrplot)## Warning: package 'corrplot' was built under R version 4.5.3## corrplot 0.95 loadedlibrary(GGally)## Warning: package 'GGally' was built under R version 4.5.3steam <- read.csv("C:/Users/HP/Downloads/steam_games_2026.csv")———————————————————
Level 1: Understanding the Data (Basic Exploration)
———————————————————
Question 1.1: What is the structure of the dataset (number of rows, columns, and data types)?
str(steam)## 'data.frame': 1000 obs. of 12 variables:
## $ AppID : int 730 2868840 3321460 3065800 3764200 2157830 1144200 1172470 1808500 2852190 ...
## $ Name : chr "Counter-Strike 2" "Slay the Spire 2" "Crimson Desert" "Marathon" ...
## $ Release_Date : chr "2012-08-21" "2026-03-05" "2026-03-19" "2026-03-05" ...
## $ Primary_Genre : chr "Action" "Indie" "Action" "Action" ...
## $ All_Tags : chr "FPS;Shooter;Multiplayer;Competitive;Action;Team-Based;eSports;Tactical;First-Person;PvP" "Strategy;Roguelike;Card Game;Deckbuilding;Co-op;Roguelike Deckbuilder;Card Battler;Turn-Based Combat;Multiplaye"| __truncated__ "Action;Open World;Singleplayer;Adventure;Combat;Exploration;Dragons;Story Rich;Action-Adventure;Third Person" "Extraction Shooter;PvP;Multiplayer;Sci-fi;Shooter;FPS;Futuristic;Action;First-Person;PvE" ...
## $ Price_USD : num 0 25 70 40 70 ...
## $ Discount_Pct : int 0 0 0 0 0 0 0 0 0 0 ...
## $ Review_Score_Pct : int 83 97 0 90 96 83 85 65 79 75 ...
## $ Total_Reviews : int 4980365 49549 0 24360 79667 2765 243810 1577 264150 1570 ...
## $ Steam_Deck_Status: chr "Unknown" "Unknown" "Unknown" "Unknown" ...
## $ Estimated_Owners : int 149410950 1486470 0 730800 2390010 82950 7314300 47310 7924500 47100 ...
## $ X24h_Peak_Players: int 1013936 0 0 0 0 0 4296 124262 0 0 ...Interpretation: This dataset of STEAM contains games data across 1000 records and 12 columns. It includes information such as game name, release date, genre, price, reviews and estimated owners.
Question 1.2: Are there any missing values in the dataset?
colSums(is.na(steam))## AppID Name Release_Date Primary_Genre
## 0 0 0 0
## All_Tags Price_USD Discount_Pct Review_Score_Pct
## 0 0 0 0
## Total_Reviews Steam_Deck_Status Estimated_Owners X24h_Peak_Players
## 0 0 0 0Interpretation: The dataset has no missing values, ensuring complete data for analysis.
———————————————————
Level 2: Data Extraction & Filtering
———————————————————
Question 2.1: Top 10 most played games
result1 <- steam %>%
arrange(desc(X24h_Peak_Players)) %>%
select(Name, Price_USD, X24h_Peak_Players)
result1[1:10, ]## Name Price_USD X24h_Peak_Players
## 1 Counter-Strike 2 0.00 1013936
## 2 Dota 2 0.00 623941
## 3 PUBG: BATTLEGROUNDS 0.00 314682
## 4 ELDEN RING NIGHTREIGN 39.99 163599
## 5 Rust 39.99 143870
## 6 Apex Legends™ 0.00 124262
## 7 Marvel Rivals 0.00 106611
## 8 DELTARUNE 24.99 83936
## 9 Call of Duty® 0.00 67419
## 10 NARAKA: BLADEPOINT 0.00 66954Interpretation: Top 10 games are counter strike, Dota 2, Pubg and more. Most of the games with high playtime are free.
Question 2.2: Top 10 highest-rated games
result2 <- steam %>%
arrange(desc(Review_Score_Pct)) %>%
select(Name, Price_USD, Review_Score_Pct, Total_Reviews)
result2[1:10, ]## Name Price_USD Review_Score_Pct
## 1 The Jackbox Party Pack 3 24.99 100
## 2 STAR WARS™ Empire at War - Gold Pack 19.99 100
## 3 Funi Raccoon Game 10.79 99
## 4 Creature Kitchen 7.99 99
## 5 Crab Champions 6.99 99
## 6 Sam & Max: Beyond Time and Space 9.99 99
## 7 Opus Magnum 19.99 99
## 8 Stardew Valley 14.99 98
## 9 DREDGE 9.99 98
## 10 Subnautica 29.99 98
## Total_Reviews
## 1 5065
## 2 34891
## 3 463
## 4 3126
## 5 25466
## 6 428
## 7 5304
## 8 846150
## 9 43759
## 10 308655Interpretation: Highest rated games are mostly paid but total number of reviews are less compared to free games.
———————————————–
Level 3: Grouping & Summarization
———————————————–
Question 3.1: How many games are free vs paid?
result3 <- steam %>%
mutate(freevspaid = ifelse(Price_USD == 0, "FREE", "PAID")) %>%
group_by(freevspaid) %>%
summarise(total_games = n())
result3## # A tibble: 2 × 2
## freevspaid total_games
## <chr> <int>
## 1 FREE 108
## 2 PAID 892Interpretation: There are 108 free and 892 paid games.
Question 3.2: Which genre earns the most? and Which genres are most popular?
result4 <- steam %>%
filter(Price_USD > 0) %>%
group_by(Primary_Genre) %>%
summarise(total_sales = sum(Price_USD), total_Reviews = sum(Total_Reviews)) %>%
arrange(desc(total_Reviews))
result4## # A tibble: 12 × 3
## Primary_Genre total_sales total_Reviews
## <chr> <dbl> <int>
## 1 Action 13343. 37090767
## 2 Indie 1237. 4234764
## 3 Adventure 1906. 3925693
## 4 RPG 1618. 3022916
## 5 Casual 986. 2716910
## 6 Simulation 1372. 1487554
## 7 Strategy 708. 1332337
## 8 Racing 671. 595802
## 9 Massively Multiplayer 102. 196181
## 10 Sports 20.0 12634
## 11 Early Access 30.0 11447
## 12 Unknown 0.49 0Interpretation: Action games are most sold and are most popular among the other generes.
Question 3.2: How many games are released each year?
steam$Release_Date <- as.Date(steam$Release_Date)
steam$year = year(steam$Release_Date)
result5 <- steam %>%
group_by(year) %>%
summarise(total_games = n())
result5## # A tibble: 22 × 2
## year total_games
## <dbl> <int>
## 1 2006 1
## 2 2007 2
## 3 2008 3
## 4 2009 6
## 5 2010 5
## 6 2011 15
## 7 2012 32
## 8 2013 17
## 9 2014 12
## 10 2015 24
## # ℹ 12 more rowsInterpretation: Most of the games were released in between 2021 to 2026.
Question 3.3: Average game owners of different genres
steam %>%
filter(Estimated_Owners > 0) %>%
group_by(Primary_Genre) %>%
summarise(avg_owners = mean(Estimated_Owners))## # A tibble: 11 × 2
## Primary_Genre avg_owners
## <chr> <dbl>
## 1 Action 2654427.
## 2 Adventure 1286517.
## 3 Casual 1482257.
## 4 Early Access 171705
## 5 Indie 2082723.
## 6 Massively Multiplayer 545468.
## 7 RPG 1374053.
## 8 Racing 1117129.
## 9 Simulation 992971.
## 10 Sports 379020
## 11 Strategy 1642166.——————————————————————-
Level 4: Feature Engineering (Creating New Insights)
——————————————————————-
Question 4.1: Creating new column Rating(1-5 stars) and game-category
steam <- steam %>%
mutate(rating = case_when(
Review_Score_Pct <= 20 ~ "*",
Review_Score_Pct > 20 & Review_Score_Pct <=40 ~ "**",
Review_Score_Pct >40 & Review_Score_Pct <=60 ~ "***",
Review_Score_Pct >60 & Review_Score_Pct <=80 ~ "****",
Review_Score_Pct >80 ~ "*****",
))
steam <- steam %>%
mutate(category = case_when(
Price_USD == 0 ~ "FREE",
Price_USD >0 & Price_USD < 25 ~ "AFFORDABLE",
Price_USD >= 25 ~ "PREMIUIM"
))
steam %>%
select(Name, Price_USD, rating, category) %>%
head(10)## Name Price_USD rating category
## 1 Counter-Strike 2 0.00 ***** FREE
## 2 Slay the Spire 2 24.99 ***** AFFORDABLE
## 3 Crimson Desert 69.99 * PREMIUIM
## 4 Marathon 39.99 ***** PREMIUIM
## 5 Resident Evil Requiem 69.99 ***** PREMIUIM
## 6 John Carpenter's Toxic Commando 39.99 ***** PREMIUIM
## 7 Ready or Not 49.99 ***** PREMIUIM
## 8 Apex Legends™ 0.00 **** FREE
## 9 ARC Raiders 39.99 **** PREMIUIM
## 10 Monster Hunter Stories 3: Twisted Reflection 69.99 **** PREMIUIMInterpretation: rating is given as 1-5 stars to the games according the review scores and games are separated into category as free, affordable and premium based on price of the games.
———————————————————
Data Visualization
———————————————————
v1
Line Chart: Games released each year
ggplot(result5, aes(x = year, y = total_games)) +
geom_line(color = "blue", linewidth = 1) +
labs(title = "Number of Games Released Each Year",
x = "Year",
y = "Number of Games")## Warning: Removed 1 row containing missing values or values outside the scale range
## (`geom_line()`).
Interpretation: There was a rise in release of games after 2020.
v2
Bar Chart: free vs paid games
ggplot(result3, aes(x = freevspaid, y = total_games, fill = freevspaid)) +
geom_bar(stat = "identity") +
labs(title = "how many games are free and paid",
x = "free vs paid",
y = "number of games")
Interpretation: most of the games are paid showing mostly premium set of games are available at steam. ## Bar Chart: Sales of the games according to different genre
ggplot(result4, aes(x = Primary_Genre, y = total_sales, fill = Primary_Genre))+
geom_bar(stat = "identity") +
labs(title = "sales of games of different genre",
x = "Genre",
y = "sales")
Interpretation: Most of the games sold are of action genre.
v3
Scatter Plot: Total games owned and played regularly
scatt <- steam %>%
mutate(owners = Estimated_Owners/100, players = X24h_Peak_Players/100, reviews = Total_Reviews/100)
ggplot(scatt, aes(x = owners, y = players)) +
geom_point() +
labs(title = "Total owners vs Total players playtime")
Interpretation: There are more owners than the playtime and there are some outliers.
Scatter Plot: Price vs Reviews
ggplot(steam, aes(x = Price_USD, y = Total_Reviews)) +
geom_point(alpha = 0.6) +
labs(title = "Price vs Reviews")
Interpretation: No strong linear relationship exists between price and reviews, suggesting that popularity is influenced more by gameplay quality and marketing rather than cost.
v4
Histogram
result6 <- steam %>%
filter(Price_USD >0)
ggplot(result6, aes(x = Price_USD)) +
geom_histogram(
aes(y = ..density..),
binwidth = 1,
fill = "#2fa",
color = "black",
alpha = 0.7
) +
geom_density(color = "blue", size = 1.5, adjust = 1.5)## Warning: Using `size` aesthetic for lines was deprecated in ggplot2 3.4.0.
## ℹ Please use `linewidth` instead.
## This warning is displayed once per session.
## Call `lifecycle::last_lifecycle_warnings()` to see where this warning was
## generated.## Warning: The dot-dot notation (`..density..`) was deprecated in ggplot2 3.4.0.
## ℹ Please use `after_stat(density)` instead.
## This warning is displayed once per session.
## Call `lifecycle::last_lifecycle_warnings()` to see where this warning was
## generated.
Interpretation: data of price is slightly right skewed
v5
Pie Chart: Show proportion of games by category (for eg.- free, affordable and premium)
result7 <- steam %>%
group_by(category) %>%
summarise(total = n())
pie(result7$total, labels = result7$category, col = c("#fc2", "#987", "#123"),
main = "Proportion of games by category")
legend("topright", legend = result7$category, fill = c("#fc2", "#987", "#123"))
Interpretation: Yellow segment indicates most of the games are affordable
Pie Chart: Show proportion of games by rating(1-5 *)
result8 <- steam %>%
group_by(rating) %>%
summarise(total = n())
pie(result8$total, labels = result8$rating, col = c("#234", "#3c2", "#986", "#097", "#fc2"),
main = "proportion of games by rating")
legend("topright", legend = result8$rating, fill = c("#234", "#3c2", "#986", "#097", "#fc2"))
Interpretation: Yellow segment indicates most games received 5 star rating
v6
Pair Plot: Relationships among price, owners, reviews and playtime
pair_data <- steam[, c("Price_USD", "Total_Reviews", "Estimated_Owners", "X24h_Peak_Players")]
ggpairs(pair_data)
v7
Box Plot: Compare reviews variability across different genres
ggplot(steam, aes(group = Primary_Genre, x = Primary_Genre, y = Review_Score_Pct, fill = Primary_Genre)) +
geom_boxplot(outlier.colour = "darkgreen", outlier.shape = 16) +
labs(title = "reviews variability across different genres",
x = "GENRE",
y = "REVIEWS") +
theme_minimal()
Interpretation: simulation and unknown genre exhibit higher medians compared to other genres having low medians and variability. Action and RPG have more outliers.
Box Plot: Compare price variability across different genres
ggplot(steam, aes(group = Primary_Genre, x = Primary_Genre, y = Price_USD, fill = Primary_Genre)) +
geom_boxplot(outlier.colour = "darkgreen", outlier.shape = 16) +
labs(title = "Price variability across different genres",
x = "GENRE",
y = "Price($USD)") +
theme_minimal()
Interpretation: Racing , simulation, action and unknown genre exhibit higher medians compared to other genres having low medians and variability. adventure and casual have some outliers.
———————————————————
Advanced Engineering
———————————————————
1.1
Simple Linear Regression
steam_model <- lm(Total_Reviews ~ Estimated_Owners ,data = steam)
summary(steam_model)## Warning in summary.lm(steam_model): essentially perfect fit: summary may be
## unreliable##
## Call:
## lm(formula = Total_Reviews ~ Estimated_Owners, data = steam)
##
## Residuals:
## Min 1Q Median 3Q Max
## -2.149e-09 -5.530e-12 -6.200e-13 1.114e-11 8.023e-11
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) -1.473e-11 2.312e-12 -6.371e+00 2.87e-10 ***
## Estimated_Owners 3.333e-02 3.488e-19 9.557e+16 < 2e-16 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 6.997e-11 on 998 degrees of freedom
## Multiple R-squared: 1, Adjusted R-squared: 1
## F-statistic: 9.134e+33 on 1 and 998 DF, p-value: < 2.2e-16predict_data <- data.frame(Estimated_Owners = c(23456, 60000, 1000, 8500))
predict(steam_model, predict_data)## 1 2 3 4
## 781.86667 2000.00000 33.33333 283.33333ggplot(scatt, aes(x = owners, y = reviews)) +
geom_point(color = "#fc2", size = 3) +
stat_smooth(method = "lm",
formula = y ~ x,
color = "blue",
size = 1.5,
se = FALSE) +
labs(title = "Estimated Owners vs Total reviews (simple linear model)",
x = "estimated owners",
y = " total reviews") +
theme_minimal()
1.2
Correlation
steam_cor <- steam[, c("Price_USD", "Review_Score_Pct", "Total_Reviews", "Estimated_Owners", "X24h_Peak_Players")]
cor_steam <- cor(steam_cor)
corrplot(cor_steam, method = "color", addCoef.col = "#111", number.cex = 0.7)
Interpretation: There is a strong positive correlation between Estimated Owners and Peak Players, indicating that games with a larger user base tend to have higher active player counts. However, price shows weak correlation with review scores, suggesting that higher cost does not guarantee better quality.