Author: Anita.x

前言：

好吧，是英文。当老师让我在全校面前参赛经验的时候 I was like

以及

（图片来源网络）

本校，是一所文科生众多的学校。多到什么程度呢？

数学竞赛班这种体制内大家赶着去的地方，不到10%的人参加，其中包括我的50%根本就听不下去。

我不是歧视，但是这事儿很难办。

写着写着我发现其实还挺简单，比正经科普简单多了（写作同学留下了不学无术的眼泪）皮就完事了。

中心思想大概是这样的：

什么？建模？优化类模型好难怎么办？

—— 不选这题就完事了

数据找不到咋办啊？

—— 你把这个指标给我筛了啊，不挺好一理由摆那里你咋不用

那这些指标摆那尴不尴尬？

——这不还有最后一题么。

至于为什么是英语？因为这是升旗仪式的稿子，要是中文我得被甲方挑到哭泣。

如果有任何人愿意看的话，凑活着看看吧。

Good morning teachers and fellow students, it’s my honor here to share some experiences on the math modeling contest in 2018. We kind of get a rewarding outcome with the design of an app “Roller Outstander”, which you can refer to the poster made by Felix Liu from 11 F. The name conveys the wishes we never expected before the contest. To be honest, our very first attempt of the “past paper” looks like this. To eliminate the confusion, our final product is like this one. So what happened?

Well, it was not really a miracle, as they are two very different models. The one for homework is more about optimization, while the one for the contest is more about the evaluation. The most effort is in design, not math, so this is a story that worth telling to everyone, including those not expecting of doing anything like math modeling like me one year ago. 

I have some special experiences with evaluation model. Before I began the training for the contest, I’ve contributed to build one in the field trip of economics in Grade 10, which some of you might be familiar. We built an index in that field research, which was actually the very basic idea of an evaluation model - building an index. They have so many similarities that I didn’t really have to study for the evaluation model. So we pick the old friend, in the real contest, and most people did. It’s a way easier, after all. 

This is the data released on the official site. Problem A was that nice and comprehensive model, while problem B is something mixed with machine learning and vague descriptions. We did want to challenge ourselves, yet we literally gave up as no one really could solve the problem. But it’s OK. We can avoid that, like the other five hundred teams. 

So the first take-away point that you might get is, you don’t really have to be comprehensive for the competition, though it might generate the feeling of regret of not knowing something awesome. 

Since the evaluation model is mostly simple, how to make it outstanding? 

This is done through the details. The data processing, the basic logic, the screening of indexes, the interpretation and application of the results - those small steps that contribute to the whole picture. Treat it as an art, even though it’s math so you won’t really believe it’s a requirement. 

Even in such a simple model, there are parts which we couldn’t deal with at first glance. Remember this beautiful thing with all the crazy idea about assessing roller coaster? The question is, where is the data? We might be able to find them all, but there are three hundred roller coasters on the list. That’s how data screening works. Our goal is to make an objective model, which is a good reason to reject them all in the ranking. They are not useless though, the design of the app is the place where we put them all. It really fits more with the common sense of assessing something recreational. 

It’s like a journey to build a model. You might never expect the destination in details, yet the scenes are beautiful enough to carry you further. You might struggle, as there are obstacles indeed. The way you go determine your outcome. After all, it’s just another way to observe the world. 

Keep learning, and explore more. Thank you. 

学无止境，写作同学立下了学习如何使用mac端计算器（matlab）的flag，是真的有趣。