UFC Fighters winrate - Exploratory Data Analysis (EDA)
Author: stiven rodriguez Course: Introduction to Data Science Assignment: EDA & Dataset Analysis Dataset Source: Kaggle – UFC Fighters' Statistics Dataset Dataset Size: ~4,112 rows × 18 column
Project Goal
The Project Goal was answer the question : What are the defining attributes—both physiological (physical) and strategic (fighting style)—that characterize the ultimate mixed martial arts fighter?
Dataset Overview
The dataset contains information about UFC Fighters winrate with the following key columns:
- name
- height_cm
- weight_in_kg
- stance
- Age
- reach_in_cm - Arm span length (cm)
- stance
- Strikes/Min - significant_strikes_landed_per_minute
- Strike Accuracy - significant_striking_accuracy
- Absorb/Min - significant_strikes_absorbed_per_minute
- Strike Defense - significant_strike_defence
- Takedowns/Fight - average_takedowns_landed_per_15_minutes
- TD Defense - takedown_defense
- Submissions/Fight - average_submissions_attempted_per_15_minutes
- TD Accuracy - takedown_accuracy
- total_fights
- win_rate
The final dataset is the result of systematic cleaning, validation, and filtering procedures
Data Cleaning Summary
- Selected 13 relevant columns from the original 18 columns for analysis, and subsequently engineered 3 new features.
- Handling Non-Informative & Duplicates
- Missing Value Handling, Removing Fighters with 3 or More Missing Columns
- Filtering 3 Fighters Due to Missing Height Data
- Filtering Fighters with Missing Date of Birth
- Filling Missing 'Reach' Values with the Mean
- Filling Missing 'Stance' Values with 'Unknown
- Column Renaming for Readability
Research Questions & Insights
*Q1: "Does age influence the win rate (WINRATE) of UFC fighters, and what is the distribution of ages among the fighters in the dataset?
Visualization : barplot
#The graph does not include fighters over the age of 38,
Click on the image above to view it in full size
Insight:
- My conclusions are that age is highly significant for successfully entering the UFC (meaning fighters need many wins on the way, which often happens around age 30+). However, within the UFC, there is no correlation between a fighter's age and their win rate (WINRATE)
*Q2: Is there a correlation between a physical variable or fighting style with the Win rate?
Visualization: Correlation Heatmap of Variables
Click on the image above to view it in full size
Insight:
- The analysis suggests that there is no strong predictive variable for success, given the consistently weak correlations found across all features.
- Despite this, the maximum correlation observed was specifically associated with measures of a fighter's 'aggressive behavior during the bout.'
*Q3: Given that a fighter possesses all the 'aggressive' attributes with a positive correlation to victory, are these traits sufficient,
on their own, to reliably predict or guarantee a high Win Rate? Visualization: Correlation Heatmap of Variables
Click on the image above to view it in full size
Insight:
- The conclusion is negative. While 'aggressive' attributes demonstrated the highest (positive) correlation with victory, these traits are not sufficient on their own
- to reliably predict or guarantee a high Win Rate. This determination is based on the finding that the overall correlation remains extremely low,
- even when considering the three most highly correlated variables—a fact visually evident in the data you referenced
Final Conclusions
Main Research Question: What are the defining attributes—both physiological (physical) and strategic
- (fighting style)—that characterize the ultimate mixed martial arts fighter?
Answer: My analysis indicates that no single variable with a strong positive correlation can be identified or
- relied upon to define the ultimate mixed martial arts fighter. Nevertheless, the data does suggest that a
- fighter's aggressive behavior during a bout is a key strategic factor that statistically elevates the probability of achieving a victory
Key Findings:
Physiological Attributes: Body measurements and physical characteristics (such as Height, Reach, or Leg length) show no significant correlation with a fighter's
probability of victory (WINRATE). This suggests that success in the UFC is not determined by inherent physical size.
Strategic Aggression: Successful Takedowns (control metrics) and Strikes Accuracy (offensive volume metrics) were found to be the variables with the highest positive
correlation with victory among all examined features.
Predictive Limitation: Despite being the top-correlated factors, the relationship between these aggressive metrics and the final outcome is still relatively weak.
Therefore, while Takedowns and Strike Accuracy are important strategic advantages, they are not sufficient on their own to reliably predict or guarantee a high Win Rate.
Overall Insights:
- There is no single parameter that can be truly relied upon to predict victory.