 Details
 Parent Category: Programming Assignments' Solutions
We Helped With This R Studio Programming Homework: Have A Similar One?
Category  Programming 

Subject  R  R Studio 
Difficulty  College 
Status  Solved 
More Info  Assignment Statistics 
Short Assignment Requirements
Assignment Description
BUS 473
Forecasting Contest – Alumni Donations
Due November 8
The object of the project is to forecast IIT alumni donations to the annual giving campaign in Fall of 2014, using data from past donation history, demographics and information about the ask strings in mailed solicitations.
You may build a forecasting model using the dataset in the CSV file donations on Blackboard. The dataset contains records of the donations returned by each alumnus during the period from September 2014 to January 2015. Donations are categorized by which campaign the donor responded to (Mail, Phone or website).
You are to provide separate equations to predict both the probability of donation (discrete choice) and the amount of each donation given based on the individual donation history and information provided about the ask amounts sent to that donor.
I will apply your formulas to a holdout sample randomly drawn from the original dataset (the data you are given will not include any of these individual).
Donation Probabilities will be evaluated based on a likelihood model. I will apply your formula to the holdout data to get expected donation probabilities. The likelihood for each donor is your probability (p) if they donated and 1p if they did not. I will multiply the likelihoods across the holdout sample to get a score. Highest likelihood wins. The predictions for contribution amounts will be scored based on Mean Absolute Deviation. The team with the lowest MAD, mean absolute deviation, will win.
You will want to construct your own holdout samples to do model selection for this assignment,
Teams will be ranked in order of combined scores, and graded accordingly.
Means for the dataset donations are below
Variable  Mean  Std Dev  Minimum  Maximum  N  






Note: This data is confidential to the university. Please do not share it with anyone outside the class and remove from your storage when the assignment is completed.
Assignment Description
Key for Donations Database
The Database contains response data from a mailing to 2081 IIT alumni.
Each alumnus received a mailed request for a donation that included specific suggested donations amounts, listed as an “ask string” in order, high, medium and low (Left, Middle, Right in the Dataset). For each recipient the ask amounts were customized based on the largest previous gift from that donor. In addition recipients were randomly divided into four treatment groups labeled A,B,C, D and the formula for computing the ask amount varied by treatment (A most aggressive – largest ask amounts – D lowest ask amounts). The purpose of the study was to determine if changing the ask amounts affected either the probability or amount of donations. You job is to provide the models to predict whether a recipient made a donation and how much they gave based on their previous history, demographic information and ask amounts/treatment. Remember the first 4 fields are not usable as predictors in your model. They are the dependent variables you want to predict (plus the unique id number)
ID – unique id no for recipient
Mail – dummy indicated whether a mail donation was received
Gifts – dummy indicating whether any donation was received (includes Phonathon pledges)
CnGf_1_Amount – total amount of gifts, if any
TestGroup1 – treatment assignment
Gender (Male/Female)
St – State of Residence
Type Graduate or Undergraduate degree
Class – year degree was awarded
Left – largest ask amount in solicitation
Middle – middle ask amount
Right lowest ask amount
Largest – amount of largest previous gift (5 years)
Last  amount of most recent gift
LGDate – Last Gift Date – when most recent donation was recieved
FY13 amount donated in 2013
FY14 amount donated in 2014
Calculation (Random variable to separate training and holdout samples – should be 1 for training sample)