Report[ENG].txt

Introduction:

This project aims to analyze the games played in the American Basketball League (NBA) between 2016 and 2018.
based on databases where various metrics are recorded. Data in the database is preprocessed 
and data processing steps were applied. In this project, prediction of match scores 
a classification performance was evaluated.

Data:
There are 4920 samples in total.
Each sample contains 123 features. Therefore, each file contains a table with 4920 x 123 entries. 
2460 samples were collected from 2016-2017 matches and 2460 samples were collected from 2017-2018 matches.
The dataset is provided as 2016-2017.csv and 2017-2018.csv. The first line of the files contains the sample names
while the other rows indicate that the metrics of the relevant instances have been collected.

Objective:
This project aims to achieve the highest percentage of correct classifications.
For this purpose, the target variable was selected and predicted based on other metrics.
Then the accuracy and cross-validation of various classification algorithms were calculated. 

Classification Algorithms:

- Random Forest 
- SVC
- KNeighbors 
- Decision Tree 

Performance Measures:
Sensitivity, specificity and AUC were calculated as outputs of program performance.

Sensitivity: = Number of correct predictions of the first class / Total number of samples in the first class

Specificity = Number of correct guesses in the second class / Total number of samples in the second class

AUC: Area under the ROC curve.

*************************************************************************************

Random Forest Classifier:

Average Cross Validation Score: 1.0
Accuracy: 1.0
Precision: 1.0
Recall (Sensitivity): 1.0
F1 Score: 1.0
AUC (Area under the ROC Curve): 1.0


Analysis:

The Random Forest Classifier model shows excellent performance. Cross-validation score, 
accuracy, sensitivity, precision, sensitivity and F1 score are all 1.0. Furthermore, the area under the ROC curve
(AUC) is also 1.0. The model learned the features in the dataset very well and successfully made predictions.

Support Vector Classifier (SVC):

Accuracy: 0.9888211382113821
Precision: 0.9919517102615694
Recall (Sensitivity): 0.986
F1 Score: 0.9889669007021062
AUC (Area under the ROC Curve): 0.9994090909090909

Analysis:

The SVC model also shows very high performance. Accuracy, precision, sensitivity and F1 score 
has high values. The AUC value of 0.9994 indicates that the area under the ROC curve is very high.

K-Nearest Neighbors (KNN) Classifier:

Accuracy: 0.9146341463414634
Precision: 0.9176706827309237
Recall (Sensitivity): 0.914
F1 Score: 0.9158316633266533
AUC (Area under the ROC Curve): 0.9684132231404958

Analysis:

The KNN model performs well, but with a lower accuracy than the Random Forest and SVC models 
and other metrics. The AUC value of 0.9684 shows the classification ability of the model.

Decision Tree Classifier:

Accuracy: 1.0
Precision: 1.0
Recall (Sensitivity): 1.0
F1 Score: 1.0
AUC (Area under the ROC Curve): 1.0

Analysis:

The Decision Tree Classifier model achieved excellent performance in all performance metrics. Model, 
completely learned the patterns in the dataset and made excellent predictions on the training set.



Overall Analysis:

All models perform well, but Random Forest and Decision 
Tree Classifier models have arrived. These models learned the patterns in the dataset very well and were able to 
set of successful forecasts.