Cheese comes in a variety of textures and firmnesses. According to their degree of firmness, cheeses can be hard, firm, veined/blue, semi-soft, soft, or fresh. A multi-class classification model was built to determine cheese firmness based on cheese nutrition and the process used to make the cheese. The EDA shows the impact characteristics like moisture content, fat level and milk treatment on different firmness levels. Consumers, chefs, and cheese-making businesses, among other industries, can benefit from a cheese classification model.

We used the Random forest model for this classification problem. Using RandomizedSearchCV, we found the best parameters which scored the highest on the training and validation sets using the metric f1 weighted. When we compare our results to the baseline model, we got a nearly 30% improvement. The scores were as follows: 0.9546 on the training set, 0.6188 on the validation set, and 0.6373 on the test set. Since the model was trained on a limited set of data, the overall results were as expected. The model could be improved by increasing the data size to allow for enough observations for each class.