MatrixVision
MatrixVision
Confusion Matrix
A confusion matrix is a performance measurement tool used in machine learning and statistics to evaluate the accuracy of a classification model. It is a table that allows visualization of the performance of an algorithm by comparing the predicted and actual classes of a dataset.
A confusion matrix typically has four cells, representing the following:
- True Positives (TP): The number of data points correctly classified as belonging to the positive class. Instances that are correctly predicted as positive.
- True Negatives (TN): The number of data points correctly classified as belonging to the negative class. Instances that are correctly predicted as negative.
- False Positives (FP): The number of data points incorrectly classified as belonging to the positive class (actually negative). Instances that are incorrectly predicted as positive (Type I error).
- False Negatives (FN): The number of data points incorrectly classified as belonging to the negative class (actually positive). Instances that are incorrectly predicted as negative (Type II error).
A confusion matrix provides insight into the performance of a classification model, including metrics such as accuracy, precision, recall, and F1-score. It is particularly useful for evaluating the performance of binary classification models but can be extended to multi-class classification problems as well.
Scatter Plot
Confusion Matrix
MATLAB
clc;
clear all;
close all;
warning off;
clc;: Clears the command window.clear all;: Clears all variables from the workspace.close all;: Closes all figure windows.warning off;: Turns off all warnings (not generally recommended unless you are sure you want to suppress warnings).
M = readtable('M.txt');
J = readtable('J.txt');
V = readtable('V.txt');
plot(M.Var2, M.Var3);
axis equal;
figure;
plot(J.Var2, J.Var3);
axis equal;
figure;
plot(V.Var2, V.Var3);
axis equal;
M = readtable('M.txt');: Reads data from the file M.txt into a table M.J = readtable('J.txt');: Reads data from the file J.txt into a table J.V = readtable('V.txt');: Reads data from the file V.txt into a table V.plot(M.Var2, M.Var3);: Plots M.Var3 against M.Var2.axis equal;: Sets the aspect ratio of the plot to be equal.figure;: Opens a new figure window.plot(J.Var2, J.Var3);: Plots J.Var3 against J.Var2 in the new figure.axis equal;: Sets the aspect ratio of the plot to be equal for the second plot.figure;: Opens another new figure window.plot(V.Var2, V.Var3);: Plots V.Var3 against V.Var2 in the new figure.axis equal;: Sets the aspect ratio of the plot to be equal for the third plot.
durM = M.Var1(end);
durJ = J.Var1(end);
durV = V.Var1(end);
aratioM = range(M.Var3) / range(M.Var2);
aratioJ = range(J.Var3) / range(J.Var2);
aratioV = range(V.Var3) / range(V.Var2);
durM = M.Var1(end);: Assigns the last value of M.Var1 to durM.durJ = J.Var1(end);: Assigns the last value of J.Var1 to durJ.durV = V.Var1(end);: Assigns the last value of V.Var1 to durV.aratioM = range(M.Var3) / range(M.Var2);: Computes the aspect ratio of M.Var3 to M.Var2.aratioJ = range(J.Var3) / range(J.Var2);: Computes the aspect ratio of J.Var3 to J.Var2.aratioV = range(V.Var3) / range(V.Var2);: Computes the aspect ratio of V.Var3 to V.Var2.
figure;
features = readtable('Features.txt'); % Read features from file
gscatter(features.Var1, features.Var2, features.Var3);
knnmodel = fitcknn(features, 'Var3');
testdata = readtable('testdata.txt');
predictions = predict(knnmodel, testdata(:, 1:2));
Observation = [testdata(:, end) predictions];
knnmodel = fitcknn(features, 'Var3', 'NumNeighbors', 5);
predictions = predict(knnmodel, testdata(:, 1:2));
Observation = [testdata(:, end) predictions];
iscorrect = string(predictions) == string(testdata.Var3);
accuracy = sum(iscorrect) / numel(iscorrect);
misclassrate = sum(~iscorrect) / numel(iscorrect);
disp(['Accuracy: ', num2str(accuracy)]);
disp(['Misclassification Rate: ', num2str(misclassrate)]);
figure;
confusionchart(testdata.Var3, predictions);
figure;: Opens a new figure window.features = readtable('Features.txt');: Reads data from Features.txt into a table features.gscatter(features.Var1, features.Var2, features.Var3);: Creates a scatter plot of features.Var1 vs features.Var2, colored by features.Var3.knnmodel = fitcknn(features, 'Var3');: Trains a kNN classifier using features.Var1 and features.Var2 to predict features.Var3.testdata = readtable('testdata.txt');: Reads test data from testdata.txt into a table testdata.predictions = predict(knnmodel, testdata(:, 1:2));: Uses the trained knnmodel to predict Var3 values for the test data based on the first two columns of testdata.Observation = [testdata(:, end) predictions];: Combines the actual and predicted values into a matrix Observation.knnmodel = fitcknn(features, 'Var3', 'NumNeighbors', 5);: Trains a new kNN classifier with 5 neighbors specified.predictions = predict(knnmodel, testdata(:, 1:2));: Predicts again using the new model.Observation = [testdata(:, end) predictions];: Updates Observation with new predictions.iscorrect = string(predictions) == string(testdata.Var3);: Checks which predictions match the actual values.accuracy = sum(iscorrect) / numel(iscorrect);: Calculates accuracy as the proportion of correct predictions.misclassrate = sum(~iscorrect) / numel(iscorrect);: Calculates the misclassification rate.disp(['Accuracy: ', num2str(accuracy)]);: Displays the accuracy.disp(['Misclassification Rate: ', num2str(misclassrate)]);: Displays the misclassification rate.figure;: Opens another new figure window.confusionchart(testdata.Var3, predictions);: Generates a confusion chart comparing the actual vs predicted classes.