🎯 CS:GO Player Count Forecasting Project

📌 Overview

This project analyzed and forecasted the player count of Counter-Strike: Global Offensive (CS:GO) using historical data from SteamDB 📊 and statistical modeling techniques in R 🖥️. The primary focus was to assess the impact of the release of CS:GO 2 on the CS:GO player base and predict future trends.

This was a group project and was done for the Forecasting Methods course.

🎯 Motivation

The announcement and beta release of CS:GO 2 led to fluctuations in player count 📉📈, and this study aimed to quantify and predict the long-term implications of this transition. Understanding these trends was crucial for the gaming community 🎮 to anticipate changes in player engagement.

📂 Data Sources

• 🔹 SteamDB: Extracted historical player count data from 2012 to 2023.
• The data used can be found in the csv file: CSGO Lifetime Chart.csv.

⚙️ Methodology

1️⃣ Data Preprocessing

• Addressed missing values using interpolation with the average of neighboring values.
• Converted the dataset into a tsibble (time series tibble) indexed by month 🗓️.
• Dividing the data set into training and test for accurate forecasting purposes.

2️⃣ Exploratory Data Analysis

• Identified key trends and anomalies 📊.
• Noted historical events that impacted player count, such as the free-to-play transition, COVID-19, and the release of Valorant.

3️⃣ Seasonality & Differencing

• Conducted variance analysis, unit root testing, and seasonal differencing 🔍.
• Identified seasonal patterns where player counts increased during holidays and Steam sales events 🛍️.

4️⃣ Model Selection

• Considered three candidate SARIMA models:
  • 📌 SARIMA(0,1,0)(0,1,1)[12]
  • 📌 SARIMA(0,1,0)(3,1,1)[12]
  • 📌 SARIMA(0,1,0)(3,1,0)[12]
• Evaluated models based on information criteria and forecasting accuracy 🎯.
• Selected SARIMA(0,1,0)(0,1,1)[12] as the best-fit model 🏆.

5️⃣ Forecasting

• Applied the selected SARIMA model to predict the CS:GO player count from 2023 to 2024 🔮.
• Generated interval forecasts and visualized results 📉.

📊 Results & Insights

• The forecast indicated a decline in CS:GO player count due to the shift of the competitive scene to CS:GO 2.
• Seasonal trends remained significant, with peaks expected during major Steam promotional events and holidays 🎄.
• The introduction of exclusive features for CS:GO Prime members in 2021 negatively impacted player count, a trend that might continue post-CS:GO 2 release.

🔧 Tools & Libraries Used

• R: Statistical computing and modeling 📊
• tsibble: Time series data manipulation 📆
• ggplot2: Data visualization 🎨
• forecast: ARIMA/SARIMA modeling 📈

🚀 How to Run the Project

1. Installed required R packages:

   install.packages(c("tsibble", "forecast", "ggplot2"))

The R script is available in CS_GO.R, and the HTML output of the script can be found in FORECASTING_PROJECT.html.
Additionally, the code for the poster is located in FORECASTING_PROJECT.Rmd.