TU59-TU60 Group Assignment Solution

groups of 2 people. Students are responsible for the formation of groups.
Introduction & Overview
This data mining assignment requires you to analyse a data set, identify data insights, build and evaluate a number of data mining models and compare the performance of these models with previously published work on the same data set.
The data set and problem are from a well know (and discussed) problem and the data set is available for download from the
UCI ML Repository called Census Income Data Set: https://archive.ics.uci.edu/ml/datasets/census+income
The prediction task is to determine whether a person makes over 50K a year. The purpose of this data mining project is to identify the profile of people that are making more than 50k per year. Also, create different groups and categories and see which ones are less likely to make 50k per year.
The website (given above) contains the data files and attribute descriptions of the data set.
The data sets contain two files. The training set contains 32,561 instances (2/3) and the testing set contains 16,281 instances (1/3). Some of them have missing values. There is a file called “adult.names” that contains a description of the different variables. It also gives some information on the performance of different models.
The dataset is well-known in the literature and there is a large number of papers that have used this dataset over the years. These papers are listed at the button on the provided website. These papers can give you more information on the dataset but you should also be able to find more information online.
You will be required to build a number of data mining models, using Orange data mining, and compare the results that you generate with those that were generated by the original researchers.
Required Tasks
You are required to produce a report detailing your work investigating the data, building classification models, analysing the results, and comparing your results with the original findings.
The first task you should complete is a data investigation exercise, where you will document the characteristics and other information that you can determine about each Feature. Identify any data insights discovered and detail all data preparation tasks and any decisions made. This work should be completed using Orange DM software. Any additional data preparation not possible in Orange DM EM can be completed using R, Python and Go Lang. But this should be minimal compared to the use of Orange DM EM.
You will need to work through/develop a number of classification models. To do this you need to use the data mining tool used in class (Orange data mining). In this tool, you can have a number of different classification techniques and within each of these, you can modify the various parameter settings.
You will need to evaluate the results from each of the models to determine which of the models gives the best results for you. You can then compare your results with the original research and discuss the outcomes.
Since this data represents income in society and one of the goals in society is more equality. In the sense that there are no big gaps based on features like gender, origin, etc. Give some actions based on the analyzed data you think can help the government to address this issue.
The original research project used a certain number of data mining algorithms. For your assignment, you should use the algorithms that are available to you in Orange data mining.
Deliverables
You will be required to document your approach to solving and evaluating this classification problem, based on the CRISP-DM process and documentation template guide.
● Your report will probably be a maximum of 15 pages.
● Only include details and images which are important for your findings and narrative. Do not fill your report with reports/charts/etc which do not add to your discussion.
The report should clearly show your work in the following areas (similar to CRISP-DM):
● Definition of the problem
● Data Exploration, Descriptive Analytics, and Identification of data insights are important − Details of any additional data preparation (cleaning, transformations, etc), data enrichment, feature engineering, feature reduction, etc ● Details on which variables are more important to determining the output of the model.
● Details of each data mining algorithm used, the configuration settings used, etc
● Details of the evaluation and performance measures from your data mining models. Examine which one performed best, why this might have been the case and how the results compare across all the models
● Discussion of how your results compared to the results from the original research and other researchers
[minimum=3, maximum=5], and any conclusions that you can draw from this comparison
● List of actions a government can take to improve the development across different groups in society.
Submission Details
You should create one document/report containing all the material for each part of the assignment. Convert this document into a PDF. It is this PDF document that should be submitted. All images should be embedded in this document.
The maximum page limit of 15 pages for your report. Although you can create an appendix section to include more material if it is necessary.
You will need to submit your assignment on BrightSpace. You cannot submit your assignment via email.
Marking Scheme

The marking scheme for this assignment is:
● 20% Problem Definition, Descriptive Analytics, Data Insights, etc & summary of initial findings/insights
● 10% Details of any additional data preparation, data enrichment, feature engineering, feature reduction, etc
● 15% Details of each data mining algorithm used, the configuration settings used, etc
● 25% Details & Discussion of the evaluation and performance measures from your data mining models.
● 10% Discussion and comparison of your results with original and other researchers.
● 20% Provide a list of actions based on your insights the government can take to improve the development across all different groups in society
The documentation for your assignment must contain the name, student number, class, course (TU??) and year information for each student in the group. Failure to give this information will incur a 10% penalty.
This assignment can be completed individually or in groups of 2 people. Students are responsible for the formation of groups.
There will be a 10% penalty deduction will be applied for each day the assignment is late. There is no penalty for submitting early.

https://www.tudublinsu.ie/advice/exams/breachesofregulations/
Assignment Feedback
I will endeavour to mark the assignments and provide feedback via Brightspace VLE. This will consist of a mark for your assignment and a short comment on the assignment. I hope to provide this feedback before you commence Assignment B, or shortly after its commencement.