Your project requires you to
• Develop a linear multiple regression model using the sales data for an 18-month sample (July 2013 through December 2014) period but excluding the latest 6 months (All 2015 sales). Your model should include suitable dummy variables and transformations for independent variables where appropriate. This is the most important step and needs considerable thought.
• Evaluate your model against other potential specifications.
• Develop a second model based on a log-linear model. This typically uses the same independent variables as the first model, but you may need to make independent variable transformations as well.
• Test the model using sales from the latest quarter (April this year through June this year) and specifically report statistics such as MAPE, FSD, Mean and Median A/S ratio, and charts showing the distribution of A/S ratios.
• Decide which of the two models is superior (linear or log-linear) and argue whether your model would produce assessment estimates that would be within benchmark figures.
•
The Data
The data covers one Adelaide Local Government Area (LGA) but your task is to deal with one of the sub-markets that have been defined. For convenience we have define sub-markets by suburb name or two suburbs grouped together.
For 2015 the Data is in the Local Government Area (LGA) of Marion and your location options for sub-markets are as follows:
EDWARDSTOWN

Each data point represents one transaction (comparable sale) of a property in the LGA. The data is derived from the South Australian Government – Office of the Valuer General and Lands Titles Office. You should create two separate data sets (Model and Test as indicated above) that relate to your specific submarket.
• The two models should be created using data from July 2013 through December 2014 (Model data set). Data after this period should NOT be included in the model but “held out” for accuracy testing (see below).
• Data from 2015 is to be used to test the model for estimation accuracy (Test data set).
The data is available as an Excel spreadsheet from this link. Some of the data is coded and information about these codes is provided separately from this link. You do not need to use all the variables in your analysis as some of it is only descriptive.

Intermediate Report 2 – Specific Analysis (worth 40% of Final Project)
Click here to obtain a relevant template.
The following specific requirements should be followed, but you are free to add value using your ability at any step.
Part 1 – Develop a Linear Regression Model
Using the Model data set, estimate a linear multiple regression to estimate the hedonic price model. Use Price as the dependent variable against any characteristics in the data-base that you think are significant contributors to the variation in price (such as building area, land area, condition, age, style). You will need to create some dummy variables here and think carefully about which variables to include. You may also choose to create interactive variables or transformations of the independent variables.
In your writeup, provide the process you used to obtain this model (methods) and the estimation of this final model (results).
Advice: This requires you to build a multiple regression model. Use the multiple regression section for help with the basic model and then refer to the section on dummy variables for the method of dealing with categorical variables in your model. Your text book and workshops have some useful examples to assist with the interpretation of the models. The reading material will also assist you as will the excel practicals. The development of the model and selection of variables is an important step in the project. You will likely want to test a few specifications, and will be assessed on your ability to choose the “best” model.
Part 2 – Re-estimate the model I as a non-linear transformation regression model
The suggested strategy is to estimate the model using an exponential transformation. To do this you would calculate the natural log (Ln) of the sale price. You then need to find the exponents of the coefficients to provide a proper interpretation of the model.
In your writeup, provide the process you used to obtain this model (method) – this should be very short – and the estimation of this final model (results).
Advice: You should refer to the material on non-linear transformations, in-particular the use of the exponential transformation.
Part 3 – Accuracy Assessment
Perform “accuracy” evaluation tests for the model on the Test dataset. You should apply both models from step 1 and 2 which used the Model data set to the 2015 data that was withheld (the Test dataset). In particular you should calculate and report the
• MAPE
• FSD
• Mean of A/S ratios
• Median of A/S ratios
• COV of A/S satios
• Plots of the AS ratio distribution.
In your writeup, provide these accuracy results.
Advice: This testing uses two different approaches. The FSD and MAPE are widely used methods for evaluating forecast and predictions as simple errors. The use of A/S ratios is very specific to property applications and is the basis of ratio studies. The IAAO has an international standard for ratio studies and the calculations should be based on these. These methods are widely used for testing the accuracy of AVM’s and specifically for mass appraisal. You should refer to the material in the study guide from the IAAO and also to the specific excel practical that provides a useful template for your evaluation.
Part 4 – Comparison of the two models
Based on the statistical and accuracy analysis you should determine which is the best model and assesses this against relevant benchmarks for accuracy.
Write-up
Your write-up should mimic the methods (Steps 1 to 2) and results (Steps 3 to 4) section of a research paper. Conciseness is very very important. We recommend you concentrate on your process in model building so that you can stay within a 2000 word maximum (aim for 1500). Do not write up every model iteration, only the process you used (the methods) to arrive at your final model structures and.interpretation of the Excel model estimations/accuracy statistics (the results).

Do you want your assignment written by the best essay experts? Then look no further. Our team of experienced writers are on standby to deliver to you a quality written paper as per your specified instructions. Order Now, and enjoy an amazing discount!!