top of page

Categorical & Numerical Predictions

Classifiers and Regression models are classical Data Science techniques used to predict outcomes. 

Classifiers are used to classify or categorize data into different classes or categories. For example, a set of measurements describing a bridge design financial transactions or customers user experience can be used to predict different outcomes that would help the business better. A bridge design may not be apt for the type of roads it will connect. A financial transaction may be suspicious, or a customer may only visit the website to take advantage of offers. 

Regression models, on the other hand, predict an actual value such as Market price, number of new customers expected to sign on, Revenue, Strength or scores describing different metrics.

From a business perspective, Categories are more helpful as they tend to be favourable with fewer residuals than Numerical predictions. However, this greatly depends on the business and its risk tolerance. For example, predicting a 'Risk/No Risk' profile may be more accurate than predicting 'High Risk/Medium Risk/Low Risk/No Risk' since the higher the granularity, the smaller the error threshold. Nevertheless, some businesses have a low tolerance for errors, and a high accuracy on the granular predictions would be desirable. 

Understanding the strengths and weaknesses of each model with respect to the Business is crucial in delivering a trustworthy Data Science model with confidence.

For more discussions on these models, please Contact Me using the buttons below. 

bottom of page