- Homework 1 contains the very fundamentals of how a machine learning problem is to be solved.
- It majorly contains data analysis and preprocessing, generating a correlation matrix and heatmap of the same.
- Later, there is a pair plot for every field with respect to the count(the result to be predicted).
- Thorough analysis of which field is useful and which is not
- Prediction using method is better for the given data MLE or MAP.
- Comments on Model Performance.
- The process included splitting the given data into training and validation.
- Calculated parameters like mean and variance of the data for the probability density function(Gaussian in this case) .
- Predicted the labels and compared with the true value.
- Implemented K nearest neighbours algorithm from scratch and made the confusion matrix
- Implemented MDS, ISOMAP and LLE from scratch.
- Visualized the clusters, the half moon dataset and the swiss roll dataset in the smaller dimensional space.
- Compared which methods work best for which dataset.