The goal of this class is to help students understand how to analyze and interpret data. Develop the ability to predict the possible outcome by observing the behavior of the database. Traditional programming class is only interested in how to write a good program, instead of emphasizing how to think as the computer and manipulate the computer work properly. The final exam is a good example to show achievement of this goal. The instructor gave students toy frogs, which can jump consecutively in the same position. The question was to draw and explain the possible mechanical design of the frog. In order sessions the instructor will periodically reward students with visual challenges. Students are to interpret the possible system and predict the behavior represented by the data. These brainstorming challenges help us observe the system and give our own explanation.
Lesson Learned
In order to understand behavior of the system, the learning of the procedure to build model from the original data with existing software is necessary. (see Graph I) First, Filter the data from the complex information. Due to the lack of structure of the original data, it is hard to use original data for calculation. AWK is a simple but powerful software that can deal with this issue. It can also search using pattern tool and has the capability to rearrange the data. Second, Calculate the data. Two software are introduced in this step, SQLite and Octave. SQLite can cope with basic problems. Octave, like Metlab, can calculate the complex matrix. Third, Plot the calculated data. GNplot is a clear software for plotting scientific data, not like Excel which is heavy with lots of unused information during the calculation. The attention to details for presenting the findings using GNplot was especially important. Lastly, Netlogo is a well-designed software to stimulate the behavior of the models.
Graph 1: The procedure to build the model
 |  |
Ruben Huele(Instructor)
|
final presentation
|
The final presentation is to analyze the data from United Nations Framework Convention on Climate Change (UNFCCC). The data contains the complete green house gas emission from different sectors in Europe countries, and the time of the data is from 1900 to 2008. Our goal is to find some strange problems from the data, like the abrupt drop in the sector, and the conversion factor (global warming potential) effected on the total emission. The judges for the final presentation are composed of the professors from the Industrial Ecology and the researchers from the consultant company. Most of the students do not have the experience for writing the program, and worried about this assignment. However, the professor significantly motivated students. He discusses with the students to adjust the way of teaching and optimized the benefit for the education. In the end of the class, students find some valuable results and are not afraid of writing program anymore. The researchers bring these results back to the company and discussed with colleagues.
The findings from the IE students (The complete result is constructed by the professor.)
http://wiki.tudelft.nl/bin/view/Education/IndustrialEcology/FcccAtlas
My research is about the emission of the chemical industry in Ireland. The green house gas (GHG) suddenly dropped down to zero in 2003. After intensive research, I found the explanation of the investigation report from the Ireland government. The chemical industry is defined for the agriculture related chemical factories, such as the factories for the ammonia. The government owned both factories, and due to the economic difficulties they closed down in 2003. All the fertilizer had to be imported from other counties.
Link: http://wiki.tudelft.nl/bin/view/Education/IndustrialEcology/FcccHan.
Another interested finding is the UNFCCC data itself: 14% of the data is zero, and significant amount of data is confidential.
Link:http://wiki.tudelft.nl/bin/view/Education/IndustrialEcology/FcccNoortjeWorkingPage.
The analysis of the carbon capture by the forest with the two criteria.
Link: http://wiki.tudelft.nl/bin/view/Education/IndustrialEcology/FcccPietro .
It is a very good start for the programming. I know how to start the analysis from a chaos data, and analyze the possible key questions. I can present the finding clearly with the graph. In addition, I relate to passion of the practical issues from the professor. I expect my thesis will also follow current real issues.
--
課堂教學的目的
這一門課主要的目的是希望學生在學習之後，能夠藉由資料看出系統的行為，或是由系統的行為看出資料可能的形式，這和計算機概論不相同，舉期末考的例子來說，其中一題考題是老師發下一個玩具青蛙，它會在持續原地翻滾，要求大家解釋可能躲在青蛙肚子裡面的設計，如何讓青蛙可以維持原地翻滾的行為。另外在課堂上，老師也會定期提供有獎徵答的題目，讓大家按照提供的圖表，解釋這個圖表所代表的行為。
學習的內容
為了從資料看出系統行為，從拿到資料到建立模型有一連串的程續（如下圖），老師所教的就是資料處理的概念與軟體的使用。首先是從繁雜的資訊過濾出需要的資料，通常向外界所取得的資訊並無均一的規則或是不必要的資訊太多，無法直接套用到已經建立的系統，AWK是一套單純好用的軟體，可以利用搜尋的 pattern指令，找出需要的資訊，或是重新安排資料的排列規則；下一步是將有規則的資訊做運算，課堂上主要使用SQLite，另外稍微提到 Octave能夠處理複雜運算（相當於MetLab）；最後是將運算後的結果繪製成圖表，我們使用GNplot，並且強調繪圖需要注意的細節。另外一套軟體 NetLogo，功能是將所得到的資料建立模型，進行系統行為的模擬。
流程圖
| |
指導老師
|
期末報告
|
期末報告是以United Nations Framework Convention on Climate Change (UNFCCC)歐盟各國提供各行各業（sector）所產生溫室氣體排放量為基礎，資料時間包含從1990年到2008年，分析資料可能的有的問題或是現象，例如某國家的某產業的溫室氣體突然降為零是否提供假資料、不同溫室氣體的轉換係數（global warming potential）是否影響總量的統計。海報競賽是期末報告的形式，評分的成員包括系上教授與環境顧問公司的研究員。學期剛開始，許多人都沒有寫程式的經驗，對於期末要如何呈現更是一頭霧水，但是授課老師能夠鼓舞學生，課堂間也討論合適的教學方式與準備報告的問題，期末時同學對寫程式建立模型也不再感到害怕，最後報告同學發現許多值得注意的結果，部份題目也被顧問公司研究員帶回公司討論。
期末報告初步的分析結果可以參考（完整的資料，老師還在整理中）http://wiki.tudelft.nl/bin/view/Education/IndustrialEcology/FcccAtlas
我的期末報告內容是發現愛爾蘭化學產業的溫室氣體排放在2003年突然歸零，出人意外發現經過一番搜索之後，在愛爾蘭政府的官方網站公佈的調查報告得知，化學產業所代表的是關於氨肥以及其它相關農業的化學工廠的排放，因為在愛爾蘭相關的兩間工廠為公部門擁有，在考量經濟效益之後，工廠在2003年關閉，農業所需肥料就由國外進口。連結 http://wiki.tudelft.nl/bin/view/Education/IndustrialEcology/FcccHan。其他有趣的結果例如在所有國家的溫室氣體資料裡，14%的資料是零。連結http://wiki.tudelft.nl/bin/view/Education/IndustrialEcology/FcccNoortjeWorkingPage。以不同標準看森林固碳效果，分析是否有國家浮報固碳成效。連結 http://wiki.tudelft.nl/bin/view/Education/IndustrialEcology/FcccPietro。
學習的心得
對於程式撰寫，這是一個很棒的開始，面對一大筆不知道該從何下手的混亂資料，現在知道該如何抽絲剝繭，分析可能的關鍵問題，並且以淺顯易懂的方式表達。此外也學習老師處理實際問題的熱情，期許自己未來的論文也能對實際問題進行研究。 
No comments:
Post a Comment