databaseDNA - Process

Consider the following hypothetical problem*. You are in charge of admissions for the Air Force Academy. Each year are faced with the difficult challenge of choosing the most qualified students to admit from thousands of applications you receive. Lacking the time and resources necessary to interview each candidate, you need a way to shorten the list to a more manageable number.

Having gone through the program yourself, you know that it takes a certain type of individual to make it to graduation. Utilizing information gathered from different sources you know the top two reasons candidates fail to graduate:

1) Academic. The average course load is 21 hours per semester; this leaves little time to “catch up” on subjects a student is weak in.

2) Athletic. There are many strength and endurance courses that must be completed within a specific amount of time. The first day, students are expected to be able to run a mile in less than six minutes; and it becomes more difficult from there.

Given the above, you want to select individuals who are strong in the areas of academics and athletics. Additionally, you want to select individuals who have demonstrated a capacity for leadership and who are of good character. Considering the situations these officers will find themselves in after graduation, it is important that they have the ability to lead and are trustworthy.

Among thousands of applications, each contains several pages of information. Using the criteria defined above, your first task is to decide which information is actually useful in determining a student’s worthiness. You devise the following point system:

In order to reduce the amount of data entry, you decide to exclude any application that has a zero score in any category. This cuts the number of applications to 84. However, there is only room for 24. Relying on information you learned in a college statistics course, you devise the following solution to pick the best 24.

You assign the following weights for each category: Academic will count for 60% of the total score, Athletic will count for 30%, and Leadership will count 10%. After the numbers are weighted, the score for each category will be added together to create a Total Score for each student. The 24 students with the highest scores, will represent those who are the strongest in those three categories, thus having the greatest probability for success. The following table shows the results.

Student Name	Rank	Total Score	Academic Score	Athletic Score	Leader-ship Score
Student 7	1	60,433	36,540	18,270	5,623
Student 2	2	60,270	35,700	18,480	6,090
Student 23	3	60,130	36,540	17,640	5,950
Student 54	4	60,060	35,700	18,480	5,880
Student 65	5	59,803	36,540	17,640	5,623
Student 26	6	59,640	36,540	17,640	5,460
Student 35	7	59,640	35,280	18,270	6,090
Student 67	8	59,593	35,700	18,270	5,623
Student 53	9	59,500	35,280	18,270	5,950
Student 1	10	59,290	36,960	16,380	5,950
Student 56	11	59,220	36,960	16,380	5,880
Student 37	12	59,045	36,540	16,345	6,160
Student 66	13	58,928	36,960	16,345	5,623
Student 68	14	58,870	36,540	16,870	5,460
Student 76	15	58,870	35,280	17,430	6,160
Student 55	16	58,765	36,540	16,345	5,880
Student 24	17	58,753	36,960	16,345	5,448
Student 41	18	58,753	35,700	17,430	5,623
Student 52	19	58,660	34,860	17,850	5,950
Student 78	20	58,450	34,860	17,640	5,950
Student 3	21	58,450	34,860	17,430	6,160
Student 63	22	58,380	35,280	17,430	5,670
Student 81	23	58,380	34,020	18,480	5,880
Student 27	24	58,345	36,960	16,345	5,040
Student 64	25	58,333	35,700	17,010	5,623
Student 69	26	58,170	36,960	15,750	5,460
Student 25	27	58,158	35,700	17,010	5,448
Student 74	28	58,100	35,280	16,870	5,950
Student 21	29	57,960	35,280	17,430	5,250
Student 34	30	57,820	34,020	17,850	5,950
Student 22	31	57,750	35,700	17,010	5,040
Student 50	32	57,540	33,740	17,850	5,950
Student 8	33	57,400	35,280	16,870	5,250
Student 33	34	57,365	34,860	16,345	6,160
Student 82	35	57,330	34,020	17,430	5,880
Student 77	36	57,318	34,860	17,010	5,448
Student 36	37	57,260	32,690	18,480	6,090
Student 79	38	57,155	34,860	16,345	5,950
Student 75	39	57,120	35,280	15,750	6,090
Student 15	40	56,980	32,690	18,480	5,810
Student 57	41	56,980	32,690	18,480	5,810
Student 19	42	56,980	34,020	16,870	6,090
Student 28	43	56,583	32,690	18,270	5,623
Student 61	44	56,560	34,020	16,870	5,670
Student 84	45	56,420	33,740	17,640	5,040
Student 51	46	56,315	34,020	16,345	5,950
Student 62	47	56,280	34,860	15,750	5,670
Student 80	48	56,280	34,020	16,380	5,880
Student 73	49	56,268	35,700	15,120	5,448
Student 20	50	56,233	34,860	15,750	5,623
Student 4	51	56,198	33,740	17,010	5,448
Student 83	52	56,070	34,020	17,010	5,040
Student 71	53	56,058	32,760	17,850	5,448
Student 14	54	55,930	31,500	18,270	6,160
Student 46	55	55,860	31,500	18,270	6,090
Student 5	56	55,790	32,690	17,640	5,460
Student 47	57	55,650	32,690	16,870	6,090
Student 9	58	55,580	34,020	15,750	5,810
Student 38	59	55,568	33,740	16,380	5,448
Student 29	60	55,440	32,760	16,870	5,810
Student 32	61	55,440	31,500	17,850	6,090
Student 11	62	55,230	31,500	17,850	5,880
Student 17	63	55,020	32,760	16,380	5,880
Student 42	64	54,950	32,690	17,010	5,250
Student 59	65	54,950	32,760	16,380	5,810
Student 58	66	54,845	32,690	16,345	5,810
Student 31	67	54,810	33,740	15,120	5,950
Student 16	68	54,705	32,690	16,345	5,670
Student 48	69	54,530	32,690	15,750	6,090
Student 60	70	54,530	33,740	15,120	5,670
Student 18	71	54,308	33,740	15,120	5,448
Student 6	72	54,075	32,690	16,345	5,040
Student 44	73	54,040	30,240	17,640	6,160
Student 45	74	54,005	31,500	16,345	6,160
Student 49	75	53,970	32,760	15,120	6,090
Student 10	76	53,550	32,760	15,120	5,670
Student 13	77	53,550	30,240	17,850	5,460
Student 43	78	53,410	30,240	17,010	6,160
Student 70	79	53,270	32,690	15,120	5,460
Student 40	80	52,150	30,240	16,870	5,040
Student 39	81	52,080	31,500	15,120	5,460
Student 72	82	52,068	30,240	16,380	5,448
Student 12	83	52,033	30,240	16,345	5,448
Student 30	84	51,870	30,240	15,750	5,880

After surveying the completed list, you cannot help feeling somewhat prideful. By utilizing and executing generally accepted and proven practices, you are confident that you have reduced thousands of applications to the 24 most likely to succeed. If you stop here and do not read what follows, you would be correct and your pride would be justified. However, what you are about to read will render the above obsolete.

The flaw with traditional approaches, like the one above, is that they obscure the very criteria that you deem to be most important. Your capacity to make wise decisions is limited or enhanced by the quality of your business intelligence. Where our process is different from anything else is that it allows you to actually see what is important. Let’s take another look at the same data after is has gone through our process.

Quite a bit different, isn’t it? The first thing you will probably notice is the colors. Not only are they pretty, they are very informative. They are used to indicate a particular item’s importance relative to all of the other values in a column; what we call relative importance. In this example, red cells represent best scores in a column and blue cells represent the worst. Reading the colors for Student 27, you will see a candidate who is very strong academically, weak athletically, and dead last in the leadership category—certainly not what you are looking for. Yet, using the traditional method Student 27 was ranked 24^th and actually made the cut!

In fact, there are a total of 5 candidates that should not have made the cut and 6 candidates who should have made it, but didn’t! In total that is 11 or almost half of the 24! Considering what you are trying to accomplish, this is hardly acceptable. It is also worth mentioning that for the sake of simplicity the example uses a relatively small set of data, just 84 items. As the size of the data increases the results of the traditional ranking method become even more flawed. Think of it as plane that is one degree of course; after one minute, it is not that big of a deal; after an hour, it is quite significant.

Let’s say you are a business that just purchased some demographic data on 10,000 potential new customers. Like most businesses, your resources are limited and your sales force can only target 500. Would you rather use traditional methods to select the 500, or ours? Over the past decade our process has produced results as dramatic as you have just seen for companies large and small in just about every type of industry.

Not only is our process proven, it is fast. We can transform your data in a matter of minutes, not days or weeks. You can start using that data immediately to start making better decisions. You don’t have to have to be a skilled mathematician or a computer expert. If you can understand the example above, you can benefit from our services.

*This model and information provided are randomly generated and does not represent actual information. Composite scoring and weighting model obtained from information provided at www.collegeconfidential.com and does not represent actual Air Force Academy procedures. This is a hypothetical example.