Hello everyone.
Mathorcup will start tomorrow morning, and the certification cup will start tomorrow night. I shouldn’t have to talk about what models are used for each question type, which is correct but meaningless nonsense, such as what time series, gray Forecasting, neural network, for example, what kind of evaluation problem uses analytic hierarchy process, fuzzy comprehensive evaluation, are these right? right. is that useful? useless. Who doesn't know the method? The point is how to apply it.
Take a look at the award rules issued by the National Competition Jury last year:
Aren't the "copies of general models" and "routine superficial articles"
referring to some general model selection diagrams that have been transferred from no one knows where and have been circulated for an unknown number of hands? I hope everyone will not hit the judges' guns.
For mathorcup, after the start of the competition, I will bring suggestions on topic selection and preliminary ideas as soon as possible. After that, I will bring a complete explanation of the finished product, a nanny-level tutorial that will teach you how to do it step by step.
Everyone pay attention to this article three times in a row so as not to get lost.
For the specific explanation format, you can refer to my explanation of the E-question in the US competition:
https://link.zhihu.com/?target=https%3A//www.bilibili.com/video/BV1zY411v7L9/%3Fspm_id_from%3D333.999.0.0%26vd_source%3D7276d7392888986cd708274d725a2
Now that the competition is approaching, in order to help some modeling novice rescue, or to save some time for some already experienced icing on the cake, I have prepared some award-winning notices + paper templates + the most complete code reference package!
If you need free materials such as template code packages and follow-up complete thesis explanations, you can see the card at the bottom of this article.
Word version of thesis template:
Code reference package:
(It is a code reference package for Mathorcup, which contains reference codes for hundreds of models of various types of competition questions. There are as many as 34 data processing methods. It has been verified that basically all of them can run. For mathorcup, it is a Compatible, hope to bring you help)
Put some code:
clc
clear all
close all
%--------------------------------------------------------------------------
% 产生混沌序列
% dx/dt = sigma*(y-x)
% dy/dt = r*x - y - x*z
% dz/dt = -b*z + x*y
sigma = 16; % Lorenz 方程参数 a
b = 4; % b
r = 45.92; % c
y = [-1,0,1]; % 起始点 (1 x 3 的行向量)
h = 0.01; % 积分时间步长
k1 = 30000; % 前面的迭代点数
k2 = 5000; % 后面的迭代点数
Z = LorenzData(y,h,k1+k2,sigma,r,b);
X = Z(k1+1:end,1); % 时间序列
X = normalize_a(X,1); % 信号归一化到均值为0,振幅为1
%--------------------------------------------------------------------------
% 相关参数
t = 1; % 时延
d = 3; % 嵌入维数
n_tr = 1000; % 训练样本数
n_te = 1000; % 测试样本数
%--------------------------------------------------------------------------
% 相空间重构
X_TR = X(1:n_tr);
X_TE = X(n_tr+1:n_tr+n_te);
[XN_TR,DN_TR] = PhaSpaRecon(X_TR,t,d);
[XN_TE,DN_TE] = PhaSpaRecon(X_TE,t,d);
%--------------------------------------------------------------------------
% 训练与测试
P = XN_TR;
T = DN_TR;
spread = 1; % 此值越大,覆盖的函数值就大(默认为1)
net = newrbe(P,T,spread);
ERR1 = sim(net,XN_TR)-DN_TR;
err_mse1 = mean(ERR1.^2);
perr1 = err_mse1/var(X)
DN_PR = sim(net,XN_TE);
ERR2 = DN_PR-DN_TE;
err_mse2 = mean(ERR2.^2);
perr2 = err_mse2/var(X)
%--------------------------------------------------------------------------
% 结果做图
figure;
subplot(211);
plot(1:length(ERR2),DN_TE,'r+-',1:length(ERR2),DN_PR,'b-');
title('真实值(+)与预测值(.)')
subplot(212);
plot(ERR2,'k');
title('预测绝对误差')
Notes related to the award:
Mathematical Modeling Scoring Reference Standard
Summary (very important) 5 points
Data screening 35 points
Mathematical model 35 points
Data simulation 15 points
Judges’ overall feeling 10 points
Mathematical modeling investigation objectives
1. Application awareness: to solve practical problems , the results and conclusions should be in line with reality; models, methods, and results should be easy to understand and easy to apply; think about and deal with problems from the standpoint of the user.
2. Mathematical modeling: To solve problems with mathematical methods, there must be a mathematical model; the mathematical abstraction of the problem model, the method is universal and scientific, and is not limited to the solution of this specific problem.
3. Awareness of innovation: Modeling has its own characteristics, which is more reasonable, scientific, effective, and practical; it has more general application significance; it is not just for innovation.
Paper Review Criteria
Rationality of assumptions, logic of modeling, creativity of models, integrity of results, and clarity of presentation.
Thinking before modeling
Questions that need to be answered in the answer sheet - modeling needs to solve those problems;
how to answer the questions - in what form the results are expressed;
what key data should be listed for each question - modeling needs Calculate those key data;
whether to list one or more sets of numbers for each quantity - to calculate one or more sets of numbers.
! ! If the math method is wrong, it will definitely fail! !
Key Points in Modeling
The whole process of modeling should be clear, justified, with results and innovations;
there must be mathematical models, and different models for the same problem must be compared;
data must be analyzed and screened;
models must be combined with data and verified with data Pass.
Main work after modeling (important)Check the correctness, rationality and innovation of the model;
the correctness and rationality of the results;
whether the written expression is clear, the analysis is incisive, and the abstract is wonderful.
Regarding the writing of the article,
there may or may not be a table of contents;
the font and format are unified, and the full text is standardized and concise; the
total number of pages is within the specified number of pages, usually 20 to 30 pages;
the results of the questions should be easy for the examiners to find. The eye-catching position is an independent paragraph;
the abstract should be written at the end, and the method and results should be clearly explained;
"model establishment" and "model solution" can be combined into "model establishment and solution";
"model evaluation" and "model promotion" can be combined into " Model evaluation and generalization".
Title naming is very important.
You need to read each subtitle to know the general context of the whole paper. Set more titles to avoid large paragraphs of text without seeing a title. Each section has a clear general title. Do not exceed two paragraphs per measure.
The following title structure can be used for reference:
5 Model establishment and solution
5.1 Question 1 Model establishment (title that can reflect specific content is required)
5.1.1 XX data processing
5.1.2 Model XX principle
5.1.3 Use XX method to solve XX
5.1 .4 Results (it is best to state the conclusion)
Figures, tables, texts
5.2 Question 2 Modeling
5.2.1 XX data processing
5.2.2 ………
5.3
Question 3…
Let’s talk about these basically, everyone have a good rest, next It will be very liver for a few days
After the start of the mathorcup and the certification cup, I will analyze and start answering the competition questions as soon as possible, and finally there will be a complete original product. If you need relevant help, you can click on my personal card below: