MSDS6371-Project Solved

Ask a home buyer to describe their dream house, and they probably won't begin with the height of the basement ceiling or the proximity to an east-west railroad. But this Kaggle competition's dataset proves that much more influences price negotiations than the number of bedrooms or the presence of a white-picket fence!

With 79 explanatory variables describing (almost) every aspect of residential homes in Ames, Iowa, this competition challenges you to predict the final price of each home.

 

Read everything BEFORE you begin.

·        Use SAS or R or both for this project.

·        There are many models and algorithms in the world of data science.  For our class, we would like to continue to study multiple linear regression and know it on a very deep level.  For this reason, this project is limited to only the methods we have used in this class.  With that said, significant gains can be found through transformations, feature selection and feature creation (hint: exploring interactions.)

Your team’s objective is to conduct 2 analyses:

1)     ANALYSIS 1: Assume that Century 21 Ames (a real estate company) in Ames Iowa has commissioned you to answer a very important question with respect to their business.  Century 21 Ames only sells houses in the NAmes, Edwards and BrkSide neighborhoods and would like to simply get an estimate of how the SalePrice of the house is related to the square footage of the living area of the house (GrLIvArea) and if the SalesPrice (and its relationship to square footage) depends on which neighborhood the house is located in. Build and fit a model that will answer this question, keeping in mind that realtors prefer to talk about living area in increments of 100 sq. ft. Provide your client with the estimate (or estimates if it varies by neighborhood) as well as confidence intervals for any estimate(s) you provide. It turns out that Century 21’s leadership team has a member that has some statistical background. Therefore, make sure and provide evidence that the model assumptions are met and that any suspicious observations (outliers / influential observations) have been identified and addressed. Finally, of course, provide your client with a well written conclusion that quantifies the relationship between living area and sale price with respect to these three neighborhoods. Remember that the company is only concerned with the three neighborhoods they sell in. 

 

2)     ANALYSIS 2: Build the most predictive model for sales prices of homes in all of Ames Iowa.  This includes all neighborhoods. Your group is limited to only the techniques we have learned in 6371 (no random forests or other methods we have not yet covered).  Specifically, you should produce 4 models: one from forward selection, one from backwards elimination, one from stepwise selection, and one that you build custom.  The custom model could be one of the three preceding models or one that you build by adding or subtracting variables at your will.  Generate an adjusted R2, CV Press and Kaggle Score for each of these models and clearly describe which model you feel is the best in terms of being able to predict future sale prices of homes in Ames, Iowa.  In your paper, please include a table similar to the one below.  The group with the lowest public Kaggle score will receive an extra 3 bonus points on the final exam!  

Quick note on Kaggle completion:  We only have one course under our belts so far (almost), but you can compete in this competition with the tools you have now (top 40th percentile or better!). After your next course (6372), you will really be able to do well (top 25th percentile or better!). With these skills as well as the skills you pick up in Data Mining and Quantifying the World, you will be able to compete with anyone!

 

Predictive Models
Adjusted R2
CV PRESS
Kaggle Score
Forward  
.89
1272
.721
Backward
.78
1590
.945
Stepwise
.81
2001
.888
CUSTOM
.87
900
.2345
 

NOTE 1: ALL ANALYSES MUST BE DONE IN SAS or R and all code must be placed in the appendix.  Part of the grading process will be to run the code and verify the Kaggle score for each group.

 

Note 2: An extra 3 points on the final exam will be awarded to the team with the model with the lowest (best) Kaggle Score.  In the unlikely event of a tie the teams will split these points.  

 

 

Sample Format

Required deliverables in the complete report:

The format of your paper (headers, sections, etc.) is flexible, although it should contain the following information and should be written in your own words.  

 

Introduction

            Brief introduction to the questions of interest and the setting of the problem.  

 

Data Description

(Where did the data come from?  How big is it? How many observations?  Where can we find out more?  What are the specific variables that we need to know with respect to your analysis?)

 

Analysis Question 1:

            Restatement of Problem

 

            Build and Fit the Model

             

            Checking Assumptions

                        Residual Plots

                        Influential point analysis (Cook’s D and Leverage)

                        Make sure to address each assumption.

 

            Comparing Competing Models

                        Adj R2  

                        Internal CV Press  

            

            Parameters

                        Estimates

                        Interpretation

                        Confidence Intervals

 

Conclusion

            A short summary of the analysis.

            

 

 

Analysis Question 2

Restatement of Problem

 

Model Selection

                        Type of Selection

                                    Stepwise

Forward

Backward

Custom                         

 

                        Checking Assumptions

                                    Residual Plots

                                    Influential point analysis (Cook’s D and Leverage)

                                    Make sure to address each assumption

 

                        Comparing Competing Models

                                    Adj R2   

                                    Internal CV Press   

                                    Kaggle Score

 

                        Conclusion: A short summary of the analysis.  

            

 

Appendix

            Well commented SAS Code for Analysis 1 and 2