Using Orange 3 for Predictive Analytics

I recently came across a very interesting tool for predictive analytics using GUI. The tool is called Orange 3. I found this tool in my local instance of Anaconda and installed it out of curiosity. Though I love coding all my applications, I’m pretty sure Orange is a good desktop software for anyone who wants to do some kind of statistical or predictive analysis with out having to do much coding.

Lets have a look at how Orange works:

We can open orange3 from the Anaconda Navigator

Orange will display the following menu. Select New and start a new project by specifying a name for your project.

A New project will have a blank white surface and a list of different elements on the left hand side in the form of a sidebar. The sidebar contains all the mathematical and statistical functions needed.

Select the data input from the first tab in the sidebar. In our example we will be using the titanic dataset that comes pre-built into orange.

Once we double click the file icon, we can select which dataset to use.

After this we can now look at dragging in the other components. In our case we will now pull in the Random Forest block and other transformations like sampling.

This way we can sample and train the random forest using the sampled data. The below image shows the components used for finding the survivors from the titanic using train data from the titanic data.


We can also observe the predictions and find a confusion matrix in order to evaluate the performance of the model.

This is indeed a smart and fast way to train and predict models and is quiet easy to use. However tends to get a little boring for geeks like me. This is a great tool for beginners who are trying to understand predictive analytics and good for researchers working on small datasets and who don’t want to waste time of scripting in R or python. However when it comes to scaling up or handling huge amounts of data, other alternatives might do a better job.

For rookies, go ahead and install orange today to get started with Machine learning and predictive analysis.


Please follow and like us:

Leave a Reply

Notify of