Kajoo is a practical interface for machine learning algorithms. The aim of this application, to be able to use some machine learning algorithms practically like clustering or regression. The interface has written with Python PyQt5 and PySide2 packages. The functions in the backend of the application has written with NumPy, Pandas and Scikit-learn packages. For download click here.
- The user can open a .csv file or save a .csv file with Kajoo.
- The user can calculate descriptive statistics of any data practically and can see data types of variables.
- The user can cluster the data with Kajoo. In this tab (below figure), user can cluster only two variables. At the same time, the user can decide the cluster number with elbow graph.
- If the user want to use clustering algorithm with three variables, the user can cluster with Kajoo and can create a three dimensional dynamic graph. (You can see the figure below.) Just like the other tab, user can decide to cluster number with elbow graph. Intercalarily, the user can save the clustering algorithm results with variables in a .csv file.
- The user can calculate the correlations of variables. In addition, user can draw a scatter plot of two variables with Kajoo.
- The user can run regression algorithms with Kajoo. In this tab (below figure), there are three different regression algorithms (Linear Regression, Decision Tree Regression, Support Vector Regression). The user can adjust the training data. In result section, the user can see the coefficients of variables with metric values (like MSE, RMSE, MAE). So, the user can decide to which model is better with metrics.
Kajoo has only Turkish language support and just clustering and, some regression algorithms. There will be more options and languages in the future.