Use Weka to do the following:
A. Classification and Prediction
• Select dataset from UCI/kaggle repository* or from any other source. The first dataset should have continuous value for the class.
• Apply on dataset classification with neural networks using backpropagation algorithm. Use various learning parameters with different number of epochs and record your notices. Use also KNN algorithm (called IBK in weka) for the same problem, change K and record the results
What is expected: The researcher should use weka very well for classification. The researcher should also be able to recognize the effect of changing the parameter values and compare the obtained results through several runs.
• For dataset apply linear regression.
• Record your notices about the results
What is expected: The researcher should know how to use Weka for regression. The important part is the capability to compare the results obtained by the neural networks and KNN in the first part.
• For the dataset apply K-Means(SimpleKMeans in Weka) algorithms and record your notices. Apply any other clustering algorithm and record your notices.
What is expected: The researcher should first know how to use weka for clustering. The researcher also has to try various number of ks (number of clusters) and record the difference between various runs. The researcher is also supposed to selected some other clustering algorithm such as Expectation Maximization (EM) algorithm and compare the results with K-means algorithm.
What is expected:
- Description of your selected datasets.
- Describe each experiment using neural network with various learning parameters, including learning rate, momentum, number of epochs and number of neurons/ hidden layers. You should compare your obtained results with various parameters. Analyze your results.
- Show the results obtained by using K-Means algorithm for the dataset and apply any other clustering algorithm and compare the results, and give your analysis.
- You have to add the following to a word file:
o Used dataset
o Screenshots of the execution results with a simple description for each one.
*UCI datasets - [login to view URL]
Kaggle [login to view URL]