Machine learning comparison project

Dataset

water safety dataset link - https://www.kaggle.com/datasets/mssmartypants/water-quality

Machine learning methods

The machine learning methods that were used in this project are logistic regression, MLP classifier, and K nearest neighbour. Logistic regression uses the training set and looks at the different chemicals and how much are found within the water and if that was safe or not, to come up with probabilities of the likelihood of the new testing being safe (Sklearn.Linear_Model.LogisticRegression, n.d.) . Once the model has been trained it calculates a probability of the water being safe, the probability is determined by the levels of chemicals. It then would look back to its training data for a similar chemical levels and use that to make a prediction on whether this new data entry is safe to drink to not. The logistic regression model was processed quickly and gave high accuracy of predictions of over 80%. It performed best using less training data than the other models. To implement this method, it used a stronger regularization of 0.1 and a saga optimisation problem for faster convergence. Then the model trains and plots the training accuracy and then performs on the testing set and prints the accuracy of this.

Video demonstration

video demonstration link - https://livecoventryac-my.sharepoint.com/:v:/g/personal/mohamme23_uni_coventry_ac_uk/EcFdtiLgdCxMr0vJZXCIPOUB4kMpVAem5cUS8sWprTpF4A?e=WiNHa8

mohamme23/10240212-SM-s1

About

Resources

Stars

Watchers

Forks

Releases

Languages