Module I - Descriptive Statistics

Assume that you will be selling your house and you are trying to develop a model to predict the eventual sale price of your home. You are interested in getting as many offers as possible so you do not want to set the asking price of the house too high since that would result in fewer people considering the property. On the other hand if the asking price is too low, you might be tempted to sell the property at too low a price and not realize an appropriate profit. You have available some old data (about ten years old) which you feel is the best you can get in the time frame in which you are working. You feel that by approximately doubling the old prices you would get values that are appropriate for current home sales.

Assignment:

Pick a random sample of 50 homes from the enclosed 90 cases. Double the asking price and selling price and use this data to develop three predictive models:

1. Develop a model for predicting the asking price of a house;
2. Develop a model for predicting the selling price of a house;
3. Develop a model for predicting the difference in the asking price and selling price.

Show the steps and data for the steps you used in developing your model and discuss the "reasonableness" of the final models chosen. Specifically discuss their predictive ability as well as the inherent error in using them. Further comment on the final variables selected and whether or not they seem sensible based on your knowledge of real estate.

Hints:

1. You should not use the selling price or any variable based on the selling price in predicting the asking price since you will not know the selling price when you set the asking price.
2. You should not use the asking price or any variable based on the asking price in predicting the selling price since the asking price is chosen by the seller, sometimes competently and sometimes very inappropriately. Try to develop a model using the more "objective" variables in the data set.
3. You should not use either the asking price or the selling price or any variable based on either the asking price or selling price in modeling the difference between the asking price and selling price.
4. Feel free to compute new variables based on the variables for which you have data. For example, you might divide sqft by the number of rooms to get the average room size which might be important.
5. Make sure you plot the rough relationship between your dependent and independent variables to see if there is any need to transform any of the variables.
   
6. The variables CITY and HEATING are dummy variables just taking on the values zero and 1. Just treat them like any other predictor ( x ) variable.
7. Case 18 is a particularly expensive house. If your random sample includes this case you should give some thought to whether or not you will use it in your model since the price seems so different from the rest of the data. There may be other outliers in the data.
8. Make sure you look at the residuals from your predictive model to make sure that your residuals are approximately "random" in structure.

9) The data is contained in the EXCEL file "realest.xls" in the EXCEL files provided to you.

Price: $41.5

Solution: The downloadable solution consists of 29 pages, 1250 words and 9 charts.
Deliverable: Word Document