Case Study: House Prices Prediction using Advanced Regression

Hyper-Parameter Optimization

Choosing the optimal value of hyper-parameters, like below involves investigations:

• Learning rate $\alpha$.
• Regularization Parameter $\lambda$.
• $k$ in Nearest Neighbor Regression.
• Hyper-parameters meant for classification algorithms.

We investigate to find the best score on the training and the test sets. In the classification chapter, we will be looking into the automated techniques that give us the best values of the hyper-parameters above.

Evaluating models using Learning Curves

In Machine Learning we use the Learning Curves to assess the performance of the models as the epochs increase and while training the model. Epoch is one whole pass of the training Dataset. There are certain curves which help us know the model’s performance and choose the optimal numbers of epochs. We will look into the details of these Learning Curves in the Classification chapter.

Dealing with overfitting and underfitting

Overfitting and underfitting are the problems that a Data Scientist comes across more often and there are certain tips and tricks for dealing with them. We will look into the details in the Classification chapter.

Kaggle competition

We will be looking into the Kaggle Competition “House Prices: Advanced Regression Techniques”. We will apply the concepts that you have learned so far and build the model. Then, we will submit our model to the competition.