To prepare your subject data you must basically redo

This preview shows page 3 - 6 out of 13 pages.

To prepare your subject data you must basically redo the same steps in setting up your Source Data for the two most recent full seasons – 2017-2018 and 2016-2017. Also, delete all values from the wins (W) column and the playof columns as those are our predicted values.Refer to assignment instruction for more Details.
1.3.Part 2: Data Analysis1.3.1.Creating a Classification ModelIn this step you must create a classification model that uses your source data to predict if teams will be inthe playofs or not. Follow the same process as the prediction model, using classification tools to split the teams into the two groups.1.3.2.Comparing the ResultsCreate a chart that compares your predicted results for both the prediction and classification to the real results for the two seasons that are being predicted. Calculate the margin of error for your prediction model and the successful classification percentage. Include this in your report. 2.Grading2.1. DeliverablesRefer to Appendix A for more details.2.2. RubricItemWeightingNotesSpreadsheet Construction10%Predictive Model10%Classification Model10%Report50%Explanation of Process15%Justification of Model Choice25%Presentation of Results5%Possible Improvements5%Accuracy of Model10%Scaled mark based on relative accuracy. Quality of Presentation10%Subjective judgement on how well your methods are presented.
3.Data Analysis NBA Historical Data from 2011/2012 – 2017/2018This section uses Classification and Predictive algorithm to analyse the NBA historical Datasets.Following algorithms will be tested:k-NN predictionTree predictionRegression prediction3.1.Data Cleaning and TransformationFinal Source Data -Cleaning and Prepa3.1.1.Data CleanupWe downloaded the miscellaneous datasets from 2011/2012-2015/2016 seasons and combined them into the SourceData source and removed any win like statistics (L, PW, PL, MOV, and SRS). We downloaded the miscellaneous dataset for 2016/2017 and 207/2018 seasons and combined them into the SubjectData source and delete all values from the wins (W) column and the playofcolumns as those are our predicted values.3.1.2.Data TransformationFollowing transformation below were applied to the data. We created Lookup tables to categorize age, as well as playof qualification, and used a TeamID to ensure all Teams are the same. Additionally it allow us to be able to use numerical value if necessary. What we discovered was rather interesting. As shown below, three teams have changed their name over the period from 2012-2018.Age_LookupAge_GroupGroup_Value2323-2402525-2612727-2822828-2933030+4PlayofPlayof*10=IF(RIGHT(B122,1)="*",1,0)A lookup table of all the 30 teams revealed some insight into the following teams, which have changed their name over the period being studied. This data is important because it prevents the systems from treating the same team as two separate teams.

  • Left Quote Icon

    Student Picture

  • Left Quote Icon

    Student Picture

  • Left Quote Icon

    Student Picture