Data Mining COSC 2111/2110 Assignment 2 Neural Networks Assessment Type You can do this assignment by yourself or in a group of 2. If you are working in a group, please establish a group in As- signment 2 Group on Canvas. Submit online via Canvas → Assignments → Assignment 2. Marks are awarded for meet- ing requirements as closely as possible. Clarifications/updates may be made via announcements/relevant discussion forums. Due Date End of week 11, Monday 12th October 2020, 11:59pm Marks 40 1 Overview In this assignment you are asked to explore the use of neural networks for classification and numeric prediction (you may choose to use ‘Javanns’ or ‘MultilayerPerceptron’ in Weka). You are also asked to carry out a data mining investigation on a real-world data file. You are required to write a report on your findings. Your assignment will be assessed on desmontrated understanding of concepts, algorithms, methodology, analysis of results and conclusions. Please make sure your answers are labelled correctly with the corresponding part and sub-question numbers, to make it easier for the marker to follow. 2 Learning Outcomes This assessment relates to the following learning outcomes of the course. • CLO 1: Demonstrate advanced knowledge of data mining concepts and techniques. • CLO 2: Apply the techniques of clustering, classification, association finding, fea- ture selection and visualisation on real world data. • CLO 3: Determine whether a real world problem has a data mining solution. • CLO 4: Apply data mining software and toolkits in a range of applications. • CLO 5: Set up a data mining process for an application, including data preparation, modelling and evaluation 3 Assignment Details 3.1 Part 1: Classification with Neural Networks (12 marks) This part involves predicting the Class attribute in the following file: hypothyroid.arff in the directory: /KDrive/SEH/SCSIT/Students/Courses/COSC2111/DataMining/data/arff/UCI/ The main goal is to achieve the lowest classification error with the lowest amount of overfitting. For the neural network training runs build a table with the following headings: Run Archi- Param Train Train Epochs Test Test No tecture- eters MSE Error MSE Error 1 ii-hh-oo lr=.2 0.5 30% 500 0.6 40% 1. Describe the data preprcocessing tasks (including data encoding) that are required. How many outputs and how many inputs will there be? How do you handle nu- meric and nominal attributes? What are the normalizations requred? How do you deal with missing values (if present)? Include your data preprocessing scripts (if necessary) as an appendix (not part of the page count). 2. Develop a script (or elaborate a pre-processing procedure in Weka) to generate the necessary training, validation and test data files. How do you determine when to stop training a neural network? Include your data preparation script (if necessary) as an appendix (not part of the page count). 3. Describe how a trained neural network determines unseen test data instance’s class label (e.g., the “analyze” strategy in Javanns). 4. Assuming that no hidden layer is used, carry out 5 train and test runs for a network. Comment on the limitations of this single-layer “perceptron” network, as opposed to a network where one or more hidden layers are employed. 5. Assuming that one hidden layer is used, use Javanns (or Weka) to carry out 5 train and test runs for a network with 5 hidden nodes. Comment on the variation in the training runs and the degree of overfitting. Comment on the differences (if any) you observe in results on the networks with or without the hidden layer. 6. Experiment with different numbers of hidden nodes. What seems to be the right number of hidden nodes for this problem? 7. For the network with 5 hidden nodes, explore different combinations of learning rate and momentum. What do you conclude? 8. Compare the classification accuracy of Javanns (or Weka MultilayerPerceptron) with the classification accuracy of Weka J48. Comment on the pros and cons of employing these two classifiers for classification tasks. 9. [Optional for COSC2110] Experimenting with both Javanns and Weka Multilayer- Perceptron, what are the pros and cons of these two different software programs for neural network training? What makes you decide to choose to use either Javanns or Weka? Provide your reasoning. Report Length Up to two pages. 2 3.2 Part 2: Numeric Prediction with Neural Networks (10 marks) This part involves the following file: heart-v1.arff in the directory: /KDrive/SEH/SCSIT/Students/Courses/COSC2111/DataMining/data/arff/UCI/ The main goal is to achieve the lowest mean absolute error with the lowest amount of overfitting. The task is to predict the value of the chol attribute. Build a similar table of runs to the one in Part 1. 1. Describe the data preprcocessing tasks (including data encoding) that are required. How many outputs and how many inputs will there be? How do you handle numeric and nominal attributes? What scaling or normalization is required? Include your data preprocessing scripts (if necessary) as an appendix (not part of the page count). 2. Modify your script (or the Weka pre-processing procedure) from Part 1 to generate the necessary training, validation and test data files. Describe how you calculate the mean-absolute error. Does it require scaling of neural network inputs, and reverse scaling of the neural network outputs? Include your data preparation scripts (if needed) as an appendix (not part of the page count). 3. Assuming that no hidden layer is used, use Javanns (or Weka) to carry out 5 train and test runs for a network. Comment on the limitations of this single-layer “perceptron” network, as opposed to a network where one or more hidden layers are employed. 4. Assuming that one hidden layer is used, use Javanns (or Weka) to carry out 5 train and test runs for a network with 5 hidden nodes. Comment on the variation in the training runs and the degree of overfitting. Comment on the differences (if any) you observe in results on the networks with or without the hidden layer. 5. Experiment with different numbers of hidden nodes. What seems to be the right number of hidden nodes for this problem? 6. For the network with 5 hidden nodes, explore different combinations of learning rate and momentum. What do you conclude? 7. Perform a run with 5 hidden nodes and no validation data. Stop training when the MSE is no longer changing. Get the error on the training and test data. Comment on the degree of overfitting. 8. What are the differences between the relative-absolute error and mean-absolute error? Which one you’d prefer to use, and why? 9. Compare the mean absolute error of Javanns (or Weka MultiLayerPerceptron) with the mean absolute error of Weka M5P. Comment on the pros and cons of employing these two classifiers for numeric prediction tasks. Report Length Up to two pages. 3 3.3 Part 3: Data Mining (15 marks) This part of the assignment is concerned with the movie data file IMDB-movie-data.csv, which is in the directory: /KDrive/SEH/SCSIT/Students/Courses/COSC2111/DataMining/data/other/ The movie data was collected from the IMDb web site which claims to be “the world’s most popular and authoritative source for movie, TV and celebrity content”. It was col- lected to answer the question “How can we tell the greatness of a movie before it is released in cinema?” There is a full description at: https://www.kaggle.com/carolzhangdc/ imdb-5000-movie-dataset. IMDB-movie-data.csv has some changes from the kaggle file, mostly to make the genre information more usable. Your task is to analyse this data with appropriate classification, clustering, association finding, attribute selection and visualisation techniques selected from the Weka menus and identify any “golden nuggets” in the data. If you don’t use any of the above techniques, you need to say why. You need to provide a report for this analysis, focusing on the following two aspects: 1. Describing the strategy you adopted, your methodology, the runs you performed, any “golden nuggets” you found and your conclusions. 2. Discussing the advantages and disadvantages of each of your chosen data mining methods. Make sure you provide a rationale of your choices, and why it worked well (or not well) for discovering the “golden nuggets”. Report Length Up to two pages. 3.4 Part 4: Self-reflection (3 marks) In this task, you will need to provide a recorded video presentation (3 or 4 minutes, with no more than 5 presentation slides) of your reflection on what you have learnt from this course on Data Mining. In particular, you should focus on answering the following questions: • Have you gained much improved understanding of key data mining concepts and major techniques? What is your reflection on the journey (considering now that you have completed your assignment 2)? • What is your knowledge and understanding now in determining whether there is a data mining solution for a real-world problem? • What have you learned from doing both assignment 1 and 2, in terms of helping you extract meaningful patterns (i.e., “golden nuggets”) for a real-world data mining problem? You will need to record the presentation in either WEBM or MP4 format (using Studio in Canvas or any software of your own choices). Both the recorded video presentation and the presentation slides (PDF format) should be submitted through Canvas. 4 4 Alternative for this assignment It is possible for your group to choose to work on some other real-world data sets from the Kaggle Competition website: https://kaggle.com. You still need to complete all four parts (part 1, 2, 3 and 4 as described in Section 3), with the only difference being the data sets you choose to use. You need to consult the lecturer about this request individually to get an approval, before going ahead with it. 5 Submission Instructions You need to submit the following 3 files via Canvas: • one PDF file for the report covering Part 1 - Part 3 (note that each part has a 2-page limit, not couning the appendix where you could include your scripts developed). • one WEBM (or MP4) video file for Part 4. • one PDF file for your presentation slides for Part 4. 5.1 Pair work submission If you work as a pair, then please include a brief paragraph (at the end of your report pdf file) to describe how you two worked together (i.e., who has done what?), and specify the percentage of your contribution to the whole assignment. You may be called upon to give a quick presentation to demonstrate how each of you contributes to the solution of this assignment. 5.2 Late submission penalty After the due date, you will have 5 business days to submit your assignment as a late submission. Late submissions will incur a penalty of 10% per day. After these five days, Canvas will be closed and you will lose ALL the assignment marks. Assessment declaration: When you submit work electronically, you agree to the assessment declaration - https:// www.rmit.edu.au/students/student-essentials/assessment-and-exams/assessment/ assessment-declaration 6 Academic integrity and plagiarism (standard warning) Academic integrity is about honest presentation of your academic work. It means ac- knowledging the work of others while developing your own insights, knowledge and ideas. You should take extreme care that you have: • Acknowledged words, data, diagrams, models, frameworks and/or ideas of others you have quoted (i.e. directly copied), summarised, paraphrased, discussed or men- tioned in your assessment through the appropriate referencing methods • Provided a reference list of the publication details so your reader can locate the source if necessary. This includes material taken from Internet sites. If you do not 5 acknowledge the sources of your material, you may be accused of plagiarism because you have passed off the work and ideas of another person without appropriate referencing, as if they were your own. RMIT University treats plagiarism as a very serious offence constituting misconduct. Plagiarism covers a variety of inappropriate behaviours, including: • Failure to properly document a source • Copyright material from the internet or databases • Collusion between students For further information on our policies and procedures, please refer to the following: https://www.rmit.edu.au/students/student-essentials/rights-and-responsibilities/ academic-integrity. 7 Marking guidelines Factors contributing to the final mark will include the number of tasks attempted, the amount of exploration and demonstrated understanding of the algorithms, methodol- ogy, logical analysis, presentation of results and conclusions (see the marking rubrics in Canvas). 6
欢迎咨询51作业君
