Step $2$: Modelling

Explore using various algorithms like KNN$,$ Decision Trees, K$-$Means, etc., and adjust their parameters to see how each affects the outcomes. This step involves:

Exploring Different Methods: Test different algorithms to see which fits our data best.

Parameter Settings: Tweak the parameters for these methods. This is like adjusting the settings on your TV to get the best picture quality.

Best Classifier Identification: Choose the classifier that performs the best. Think of it as picking the best player for your team based on their performance.

It's all about trying different settings and picking the best one in a way that we can explain and stand behind morally.

Step $3$: Model Optimization

This step is about becoming an 'expert' user:

Explore Multiple Classifiers: Try out different classifiers and hyper$-$parameter settings. It's like experimenting with other ingredients and quantities to perfect a recipe.

Optimization: This process is called model optimization. We aim to find the best model settings that give us the highest score or performance.

Keep adjusting the dials $($parameters$)$ for different factors and observe which settings make the model perform the best.

Step $4$: Model Evaluation:

ROC $($Receiver Operating Characteristic$)$ Score: We will use the ROC curve representation to compare how different approaches perform.

Imagine the ROC curve as a graph showing which model better distinguishes between our classes. We can visually compare our models' performance using the ROC curve to see the best. It's like comparing scores in a game to see who wins.

With my dateset give, I tried to make highest accuracy, but the AUC score is always lower than $0\mathrm{.}7.$ Could you please give me tip about this problem?