How to Install Orange Data Mining Tool on Linux?

Orange is a powerful platform to perform data analysis and visualization, see data flow and become more productive. It provides a clean, open-source platform. It was developed by The  University of Ljubljana under the GPLv3 license.

Steps of Installation

Step 1: First of all, we will install pip and other dependencies before installing Orange Tool.

sudo apt install build-essential python3-dev python3-pip

To verify the installation, run:

pip3 –version

Step 2: Now, install the orange tool.

pip3 install orange3

Note: This command will also install various machine learning libraries and PyQt5 that may cost you additional data. 

Using Orange Tool

Run the following command on the command-line:

python3 -m Orange.canvas

Orange Widgets

These are the building blocks of data workflows of the visual programming environment. We have the following widgets in orange categorized according to their functionality.

Data

These widgets read and display data. Some common examples are:

• File: It reads the input data file and sends the dataset to its output channel.
• CSV File Import: It reads comma-separated files and sends the dataset to its output channel.
• Datasets: It retrieves selected datasets from the server and sends them to the output.
• Data Table: It receives dataset(s) in its input and presents them as a spreadsheet.

Visualize

These widgets visualize the given data through various graphs and bars. Some common examples are :

• Box Plot: It shows the distributions of attribute values.
• Distributions: It displays the value distribution of discrete or continuous attributes.
• Scatter Plot: It provides a 2-dimensional scatter plot visualization.

Model

These widgets apply machine learning algorithms to the given dataset(s). Some common examples are:

• Constant: It predicts the most frequent class or means value from the training set.
• CN2 Rule: It induces rules from data using the CN2 algorithm.
• kNN: It predicts according to the nearest training instances.
• Random Forest: It predicts using an ensemble of decision trees.

Evaluate

These widgets evaluate the result produced by the model widget. Some common examples are :

• Test and Score: It tests machine learning algorithms on data.
• Predictions: It shows models’ predictions on the data.
• Confusion Matrix: It shows proportions between the predicted and actual class.

Unsupervised

These widgets process unsupervised data. Some common examples are :

• Distance Matrix: It visualizes distance measures in a distance matrix.
• Distance Map:  It visualizes distances between objects.
• k-Means: It applies the k-Means algorithm to the data.