COMP219 - 2019 - First CA AssignmentIndividual courseworkSimple Machine Learning ModelAssessment InformationAssignment Number 1 (of 2)Weighting 10%Assignment Circulated Thursday 10 October 2019Deadline Friday November 8 2019, 15:00Submission Mode ElectronicLearning outcome assessed 2. Ability to choose, compare, and apply suitable basiclearning algorithms to simple applications;Purpose of assessment To implement machine learning algorithms on a datasetMarking criteria The marking scheme can be found in Section 3Submission necessary in order Noto satisfy Module requirements?Late Submission Penalty Standard UoL Policy.11 ObjectivesThis assignment requires you to implement and evaluate one (or multiple) simple machinelearning models on two datasets.2 Requirement and DescriptionLanguage and Platform Python (version 3.5 or above). You can use some other librariesavailable on Python platform, including numpy, scipy, scikit-learn, and matplotlib. If youintend to use libraries other than these, please consult the demonstrator or the lecturer.Dataset Please use one of the following two datasets, whose information can be found inhttps://scikit-learn.org/stable/datasets/index.html• Optical recognition of handwritten digits dataset• RCV1 datasetLearning Task You can choose either classification (preferred) or regression.Learning Model/Algorithm You may choose at least one learning algorithm from thefollowing list:• decision tree learning• naive Bayes• k nearest neighborAssignment Tasks Once you have selected a learning algorithm, you need to implementthe following functionalities:f1 provide the details of the dataset,f2 train a machine learning model by calling an algorithm from the machine learninglibraries such as scikit-learn, and save the model so that it can be called later. You cansave a model with scikit-learn built-in functionality such as https://scikit-learn.org/stable/modules/model_persistence.html.f3 implement a machine learning algorithm by yourself, train a model with the algorithm,and save the model so that it can be called later;f4 compare the train error and test error of the two models;f5 enable the user to query the saved models with e.g., an index of the test dataset.2Additional Requirements We have additional requirements that,1. the marker can run your code directly, i.e., see the results of functionalities f1, f4, andf5 by loading the saved models, without calling the training functionalities f2 and f3,and2. you need to provide clear instructions on how to train the two models, i.e., run functionalitiesf2 and f3. The instructions may be e.g., a different command or an easyway of adapting the source code.Documentation You need to write a proper document1. detailing how to run your program, including the software dependencies,2. explaining how the functionalities and additional requirements are implemented, and3. providing the details of your implementation, including e.g., the meaning of parametersand variables, the idea of your algorithm, etc.Also, the document needs to follow the guidelines in Note 1 of Section 3.Submission files Your submission should include the following files:• a file for source code,• two files for saved models, and• a document.Please see Section 4 for instructions on how to package your submission files.3 Marking CritCOMP219代做、Python编程语言调试、Python代eriaThe assignment is split in a number of steps. Every step gives you some marks.Note 1 At the beginning of the document, please include a check list indicating whetherthe below marking points have been implemented successfully. The length of the submitteddocument needs to be within 4 pages (A4 paper, 11pt font size).Note 2 The marking of a functionality will also consider the quality of coding and the qualityof documentation. A run-able implementation alone will have up to 50% of the marks.functionality f1: 20%Successfully load the dataset and display the dataset information, including the number ofdata entries, the number of classes, the number of data entries for each classes, the minimumand maximum values for each feature, and the train dataset and test dataset split.3functionality f2: 20%Successfully call library functions to train and save a model. There needs to be a correspondingsaved model in your submission.functionality f3: 40%You have an implementation of an algorithm that is able to train a model. There needs tobe a corresponding saved model in your submission. You cannot call any library which hasdirect implementation of a machine learning algorithm.functionality f4: 10%Please output the train and test errors for both models. Each model have 5%.functionality f5: 10%Allow users to query the models by changing the input. For example, you can use a variableto represent the index of the test dataset. Each model have 5%.4 Deadline and Submission Instructions• Deadline for submitting the first assignment is given at the beginning of this document.• Please submit all the files in a single compressed file with the filename00hstudentnumberi.tar00 or 00hstudentnumberi.zip00For example, “201191838.tar” or “201191838.zip” if your student number is 201191838.Submissions with other filename will not be accepted. Also, in the submission files,please do not include your name.• Submission is via VITAL Turnitin system.5 Q&AQ: What if I choose to implement k-NN which does not have a model?A: You do not have to save a model, but you need to make sure that your program canrun in less than 4 minutes (for two algorithms). You can for example take a subset of datato train. Also, you need to explain this situation (i.e., you have less files in your submissionpackage), including how long it takes for the program to run on your own machine.4Q: Shall I download the dataset from the original place e.g., UCI repository, orsklearn package ?A: Please use sklearn. Although the original repository include more data samples, we arefocused on the algorithm and would like to take an easy, and consistent, way of loading data.Q: Will I be penalised if the accuracy of my algorithm is not good?A: Accuracy is not our major concern, and our marking will not be affected by the accuracy.However, if your algorithm is not correct, you will get less marks.Q: Can I use a previous version of sklearn ?A: We recommend everyone to use the up-to-date version of sklearn (version 0.21 as atOctober, 2019). However, if you have difficulty using this version, please clearly describe inyour document which version you are working with and the reason why you cannot use thenew version.Q: I do not know how to load dataset. Can you help?A: sklearn has built-in loading function for you to call directly. Please Google to learnthis. :)5转自：http://www.3daixie.com/contents/11/3444.html
