Report[ENG].txt

Introduction:
This project aims to analyze DNA samples from human fingerprints
based on the latest research that has been discovered. Therefore, the computer
It is possible to identify the user with samples taken from their keyboard.
This task provides very strong patterns, and the recognition rate is quite high.
However, a more challenging task is to determine which hand (left or right)
to determine whether the DNA data was collected. In this project, DNA data from individuals
The classification performance of the collected clinical data was evaluated.

Data:
There are 271 samples in total (first 136 left hands, next 135 right hands).
Each sample contains 3302 features. Therefore, each file has 3302 x 271 entries.
contains a table. 136 samples were collected from the right hand and 135 from the left hand.
The data set is provided as otu.csv. The first line of the files contains the sample names
while the second line indicates whether it was collected from the left or right hand.

Objective:
This project aims to achieve the highest percentage of correct classifications.
For this purpose, various algorithms and cross-validations were tried and a neural
network algorithm was preferred. Then, with the selected features classification was performed.

Classification Algorithms:
Using a Multi-Layer Perceptron (MLP), a
classification model. In particular, the MLPClassifier class, is a multilayer neural
network-based classification algorithm and in this example is used.

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.

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

Data Reading and Editing:

Libraries: Pandas and Sklearn. preprocessing. Label encoders were used.
The data set otu.csv was read with the pd.read_csv function. Prevent DtypeWarning
The entire data set was read as a string. Data set, input (X) and output (y)
data. Labels were converted to numeric values with the LabelEncoder.

Data Set Splitting:

The data set was split into training and test sets using the train_test_split function.
divided into

Normalization and MLP Model Building:

The data was normalized with StandardScaler.
A multilayer perceptron (MLP) model was created with MLPClassifier. This model,
will use the dataset in the training process.

Model Training:

The model was trained with fit function.

Prediction and Performance Evaluation:

Predictions were made on the test set. The performance of the predictions was
evaluated, and the results were printed on the screen.

Performance Evaluation Outputs

Confusion Matrix:

 [[16 6]
 [ 9 24]]

Classification Report:

               precision    recall      f1-score    support

         0      0.64        0.73        0.68        22
         1      0.80        0.73        0.76        33

accuracy 0.73 55
macro avg 0.72 0.73 0.72 55

weighted avg 0.74 0.73 0.73 55

Accuracy: 0.7272727272727273

Sensitivity : 0.7272727272727273

Specificity : 0.7272727272727273

ROC AUC : 0.7273

According to these outputs, the overall performance of the model is good.
Accuracy and specificity are balanced and high.
This information shows that the model performs the classification task successfully.
that it has realized.