Evaluation PROBLEM: Ideal solution: Data Mining

		Course data Mining the second semester of the 2014-2015 academic year Comments: Complete content, clear thinking, The format is correct and the structure is rigorous The exposition is reasonable, the algorithm is correct, Proper analysis and accurate terminology Clear expression, fluent text, Use words accurately, no typos Theory with practice, No plagiarism Out of Note: check: Correct, complete, detailed, reasonable Half-check: partially correct, incomplete, not detailed, basic reasonable Wrong number: Wrong, missing, poor To copy a vote of veto ========================================== The ideal solution model for relevance analysis--ranking coaches Points: 1, the experimental data set can be downloaded from the Internet, you can also use the data set used to do Clementine. If downloaded from the Internet, be sure to note the data set and make a brief description. 2, can use any one or more of the analysis method ( must use at least one of the correlation analysis, clustering analysis, classification algorithm ), the data set analysis. 3, the excavation process to be detailed, to pay attention to the purpose of mining, mining process, algorithm, mining results to be visualized display, the results of the excavation to be tested and analyzed. 4, according to the results of the excavation to guide the practice of ideas, methods and so on. 5. Copying is strictly forbidden. The topic is self-prepared. 6, must use this design report template, the body A4 paper 3-5 page, double-sided printing.
	Directory A Background Introduction 2 Two Model Overview 2 Three Model Algorithm 3 Step1 of the major competition award 3 STEP2 Data preprocessing 4 Step 3 Construction weighted matrix 4 STEP4 Computing positive ideal and negative ideal solution 4 Step5 Calculating the distance between the schemes to the positive and negative ideal solutions 4 Step6 Calculate and sort the comprehensive indicator values for each scheme 5 Three Model Solving 5 Four Reference 7 Five Algorithm Code 8 A Background introduction This data set is from the United States http://www.ncaa.org/ website, complete with the data in the attachment. We can get some results by mining these data. An ideal solution based on association analysis using the relationship between the major indicators, using a specific algorithm to process the data and finally get a sort value to obtain the ranking of coaches. The specific algorithm code is shown in the appendix. We should select more indicators to establish a relatively detailed evaluation model when we evaluate coaches. We have chosen the number of years to teach, the number of participants, the winning rate and major competitions such as the number of regular champions, the number of league championships, the first round of the NCAA Championship, The number of times to enter the NCCA and get NCCA The number of champions, but these major matches reflect the same level of meaning, so we combine it into an indicator and then put it with the rest 3 Indicators constitute the indicator system for the second round of evaluation. The data format is as follows: Two Model Overview (1) First use the contribution rate as the weight, will be the regular champion, the league Championship champion, theNCCA Championship, theNCCA Four,theNCCA Champion, These five indicators are combined into one indicator (named large tournament winning rate); (2) The number of years of coaching, coaching sessions, the winning rate, the winning rate of large-scale competitions, these four indicators are ranked by the ideal solution, screening out the first ten . Among them, the weights of the ideal solution are determined by the coefficient of variation method. The ideal solution consists of six steps: (1) Standardize the processing of four indicators; (2) The normalized index is weighted and summed; (3) The positive ideal and negative ideal solution are obtained respectively; (4) The distance between the data of the group and the positive ideal solution and the negative ideal solution are obtained. (5) To seek comprehensive evaluation index; (6) ranking (select top ten ); Three Model algorithm The following is the ideal solution to find the top five coaches. Name Coaching Age General Coaching No. Winning Normalization of other factors Negative distance Positive distance Sort Mike Krzyzewski 39 1277 0.764 0.1450 0.1516 0.00747 0.9529 Jim Boeheim 38 1256 0.750 0.1381 0.1447 0.0110 0.9293 Dean Smith 36 1133 0.776 0.1452 0.1500 0.0117 0.9277 Adolph Rupp 41 1066 0.822 0.1346 0.1395 0.0151 0.9026 Lute Olson 34 1061 0.731 0.1332 0.1376 0.0190 0.8784 The following is a combination of hierarchical analysis and other algorithms to find out the top 10 coaches. Table Top Ten coaches Position 1 2 3 4 5 Coach Mike Krzyzewski Jim Boeheim Dean Smith Adolph Rupp Lute Olson Comprehensive Evaluation Index 0.9529 0.9293 0.9277 0.9026 0.8784 Position 6 7 8 9 10 Coach Bob Knight Jim Calhoun Eddie Sutton Denny Crum Roy Williams Comprehensive Evaluation Index 0.8775 0.8619 0.8292 0.8109 0.7771 Four Reference [1] Frank R. Giordano, William P. Fox, Steven b. Horton, and Maurice D. weir:a first Course in mathematical Modeling, Fou Rth Edition. [2] Matlab the Language of Technical Computing Http://www.mathworks.com/products/matlab/examples.html
		Five Algorithm code Using MATLAB software for data processing, the code is as follows % normalized processing normalization - contribution value % Chen Jinxia % time:2015.1.17 Clc Clear Load (' Second2.mat ');% Data storage Place I=1; J=[]; while (i<=2) Temp=var (A1 (:, I:i)); J=[J,TEMP]; i=i+1; End Temp=sum (j); % mean value staging place Disp (' variance '); Disp (j); % Variance Staging Place J=j./temp; Disp (' normalization '); Disp (j); % coefficient of variation temporary storage place k=1; S=[]; % normalized data storage place while (K<=50) S=[s;dot (J,A1 (k:k,:))];% weighted sum k=k+1; End Disp (s) %topsis vector Normalization processing: Suitable for virtual solution, European distance % Chen Jinxia % time:2015.1.17 CLC Clear Load (' Second2.mat ');% load data b Result=[]; % preprocessing data storage Place Jresult=[];% weighted data storage place Idearesultz=[]; % is ideal to dissolve Idearesultf=[]; % Negative Ideal resolution d=b.^2; For I=1:4 C=SQRT (SUM (d (:, i:i))); Result (:, i:i) =b (:, i:i)./C; End % data preprocessing end % coefficient of variation starts W=[]; For I=1:4 W=[W,STD (Result (:, i))/mean (Result (:, i))]; End Temp=sum (w); W=w./temp; For I=1:4 Jresult (:, I:i) =result (:, i:i). *w (1,i); End % Weighted processing complete For I=1:4 Idearesultz=[idearesultz,max (Jresult (:, i:i))]; Idearesultf=[idearesultf,min (Jresult (:, i:i))]; End % Ideal solution completed Sz=[]; % positive distance temporary storage place Szz=[]; Sf=[]; % negative distance temporary storage place Sff=[]; F=[]; For i=1:50 Sz=[sz (Jresult (i:i,:)-idearesultz). ^2]; SZZ=[SZZ;SQRT (SUM (SZ (i:i,:)))]; SF=[SF (Jresult (i:i,:)-idearesultf). ^2]; SFF=[SFF;SQRT (SUM (SF (i:i,:)))]; F=[F;SFF (i)/(SFF (i) +szz (i))];% sort Values End

Evaluation PROBLEM: Ideal solution: Data Mining