ML-Labs

Supervised Learning

In this lab we used the IRIS dataset:

• Data visiualisation to understand our dataset using pylab.scatter

• Trainig and prediction using Naive bayes

• Evaluate the classifier performance using 3 different methods

  • Sum function
  • the difference between arrays
  • Accuracy

Unsupervised learning

In this lab we used a dataset of cheese in order to classify them into different types based on their composition.

for that we used 4 algorithms

• k-means
• Hierarchical ascending classification
• Agglomerative Clustering
• Diana

Implementation of Divisive clustering (DIANA) with k-means

Algorithm :

Apply kmeans algorithm on the whole dataset with number of clusters = 2, in each resulted cluster we re-apply kmeans recursively until we get a cluster of one element. Every time we have to put the linkage data in Z matrix in order to generate the dendrogram

• Linkage data = [ index of element 1, index of element 2, distance between element1 and 2, clusters length ]