I have a historical data set for customers for a particular company. Target class being Yes/NO (Would a customer subscribe to a new product.)

I need to develop a classification approach to predict which individuals are more likely to subscribe to the product. This might involve statistical analysis and model selection. What can be the possible steps that I can incorporate in the mining process, what kind of classifiers can I use, would really appreciate a step by step brief.

Ideally, I want to use orange data mining suite for the same.


Check out a video Getting Started with Orange 06: Making Predictions to see how to set up a workflow for classification. You can replace classification tree with any other more accurate method of prediction, like Logistic regression or Random Forest.

Here is an example workflow that you can use: it reads the class-labeled input data (File - Training Data), builds a logistic regression model and then uses it in Predictions widget on your data to be used for predictions (File - Data for Prediction). Confusion Matrix displays a matrix of true and false predictions and selected data from this table (say, all mispredicted data instances) are shown in the Data Table.

enter image description here


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.