Allisona commited on
Commit
14e2cbe
·
verified ·
1 Parent(s): 6ca3646

Delete README.md

Browse files
Files changed (1) hide show
  1. README.md +0 -66
README.md DELETED
@@ -1,66 +0,0 @@
1
- ---
2
- title: UFC Fight Predictor
3
- emoji: 🥊
4
- colorFrom: blue
5
- colorTo: indigo
6
- sdk: gradio
7
- sdk_version: 4.19.2
8
- app_file: app.py
9
- pinned: false
10
- license: mit
11
- ---
12
-
13
- # UFC Fight Predictor
14
-
15
- A machine learning application that predicts UFC fight outcomes based on fighter statistics.
16
-
17
- ## Model Performance
18
-
19
- - **Accuracy**: 98.9%
20
- - **Best Algorithm**: Logistic Regression
21
- - **Dataset**: 7,417 UFC fights
22
- - **Features**: 25 statistical indicators
23
-
24
- ## Features Analyzed
25
-
26
- - Knockdown differences between fighters
27
- - Significant strike statistics
28
- - Takedown attempts and success rates
29
- - Submission attempts
30
- - Striking accuracy
31
- - Weight class
32
- - Fight round
33
- - Method of victory encoding
34
-
35
- ## How to Use
36
-
37
- 1. **Enter fighter statistics** for both competitors
38
- 2. **Select weight class** from dropdown
39
- 3. **Adjust method parameter** (0-10 scale)
40
- 4. **Click Submit** to get prediction
41
-
42
- ## Output Information
43
-
44
- The model returns:
45
- - Predicted winner
46
- - Confidence percentage
47
- - Individual probabilities for each fighter
48
- - Detailed analysis
49
-
50
- ## Technical Details
51
-
52
- Built with:
53
- - Scikit-learn for machine learning
54
- - Gradio for web interface
55
- - Pandas for data processing
56
- - NumPy for numerical operations
57
-
58
- ## Model Validation
59
-
60
- The model was validated using:
61
- - Train/test split (80/20)
62
- - Multiple algorithm comparison
63
- - Cross-validation techniques
64
- - Comprehensive metrics (Accuracy, AUC, F1-Score)
65
-
66
- *For educational and demonstration purposes.*