Machine Learning & Advanced Python

Machine Learning paper

In this module, I was given the task of using Known Machine Learning models to predict a set of outcomes of my choosing.

I chose to use a set of algorithms to classify good wine and bad wine, based on a variety of wine features.

My chosen algorithms were Logistic Regression, Random Forest Classifier and Gradient Boosting Classifier.

After choosing my dataset and 3 models, I got to work, tuning the hyperparameters and improving the learning accuracy, little by little.

I reached a peak accuracy of 64% using the Random Forest Classifier, concluding that I needed more data, more variables and a more powerful computer to detect all of the relationships between the features that affect wine quality. However, this model was able to uncover some of the complex, non-linear relationships between the features and quality variable, which aided greatly in obtaining the 64% accuracy figure.

Here is the list of features I used, complete with their correlation to the target variable, Wine Quality:

Check out the code I used for this classification task on Google Collab.