Difference between revisions of "Team:NCTU Formosa/Collaborations"

Line 506: Line 506:
 
     </tr>
 
     </tr>
 
     <tr style="background:#FFDDAA;">
 
     <tr style="background:#FFDDAA;">
−
           <td  height="40">Type</td>
+
           <td  height="30">Type</td>
 
           <td class="table_content">Open Source</td>
 
           <td class="table_content">Open Source</td>
 
           <td class="table_content">Query Website</td>
 
           <td class="table_content">Query Website</td>
 
     </tr>
 
     </tr>
 
     <tr>
 
     <tr>
−
           <td height="40">Duration</td>
+
           <td height="30">Duration</td>
 
           <td class="table_content">Collected every ten days</td>
 
           <td class="table_content">Collected every ten days</td>
 
           <td class="table_content">Collected every day</td>
 
           <td class="table_content">Collected every day</td>
 
     </tr>
 
     </tr>
 
     <tr style="background:#FFDDAA;">
 
     <tr style="background:#FFDDAA;">
−
           <td width="40" height="200">Data Type</td>
+
           <td height="30">Data Type</td>
 
           <td class="table_content">Number</td>
 
           <td class="table_content">Number</td>
 
           <td class="table_content">Temperature, Rainfall</td>
 
           <td class="table_content">Temperature, Rainfall</td>
 
     </tr>
 
     </tr>
 
     <tr >
 
     <tr >
−
           <td height="40">Source</td>
+
           <td height="30">Source</td>
−
           <td class="table_content">Taiwan Agricultural Research Institute</td>
+
           <td class="table_content"><a href="https://sheethub.com/data.gov.tw/%E8%94%AC%E6%9E%9C%E9%87%8D%E8%A6%81%E5%AE%B3%E8%9F%B2%E9%98%B2%E7%96%AB%E6%97%AC%E5%A0%B1" style="color:#44E287;">Taiwan Agricultural Research Institute</a></td>
−
          <td class="table_content">CWB Observation Data Inquire System</td>
+
           <td class="table_content"><a href="e-service.cwb.gov.tw/HistoryDataQuery/index.jsp" style="color:#44E287;">CWB Observation Data Inquire System</a></td>
−
    </tr>
+
−
    <tr style="background:#FFDDAA;">
+
−
          <td height="40">Website</td>
+
−
          <td class="table_content"><a href=""></a></td>
+
−
           <td class="table_content"><a href="e-service.cwb.gov.tw/HistoryDataQuery/index.jsp">CWB Observation Data Inquire System</a></td>
+
 
     </tr>
 
     </tr>
 +
  
 
   </table>
 
   </table>

Revision as of 13:27, 15 October 2016

Collaborations

Collaborations among iGEM teams are one of the essential parts in Human Practices, for we have become companions with other iGEMers because of the reciprocal interactions. This year, we have done some collaborations including experience assistance, mentoring a new team.

TaiwanPP_FTP

Overview

TaiwanPP_FTP is a software collaboration between NYMU-Taipei and NCTU_Formosa. As the teams that both care about agricultural issues, we collaborated with NYMU-Taipei iGEM team to create innovative software that can predict the tendency of pest prospering shortly. The software will able to give detailed information of farm condition and give advice for farmers to prepare earlier before the threat.

We built up a prediction model by using the most popular way called machine learning for a variety insects. Through training a computer with a great amount of data, the computer will be able to analyze the relationship between input conditions and the outcomes. Finally, it can provide an accurate prediction which is the useful information for farm owners.

The Aim of TaiwanPP_FTP

We hope this tool could be user-friendly. All the information could be accessed through the Internet freely. It is remarkable if users could use the existing data to recreate another great application for the world. The product was a convenient web page that will show the predicted number of pests in the following three days of each observation station in Taiwan. All the source code and data are already on GitHub.

The Process of Collaboration

This software is the cooperated product of National Chiao Tung University and National Yang-Ming University. The member of NYMU first developed this idea. The member of NYMU was responsible for developing the software including raw coding for data collection, data processing, prediction voting establishment and web development, while the member of NCTU was acting as a consultant and advisor that refined the code and debugged including stuff such as streamlining the program, database connection. During the cooperation, we had several discussions through video chats on Facebook and had discussions in person. Finally, we made it an open source software and collaborated on GiHub.

How does it work

  • Objective:
          By inputing weather data of the past few days, users can get the population size of pests. The types of pest include Bactrocera dorsalis, Bactrocera cucurbatae, Spodoptera litura, and Spodoptera exgua Hiibner.
  • Data sources:
    • Data Pests Weather
    Type Open Source Query Website
    Duration Collected every ten days Collected every day
    Data Type Number Temperature, Rainfall
    Source Taiwan Agricultural Research Institute CWB Observation Data Inquire System
  • Hypothesis:
          According to a great amount of research paper, the weather condition influences the maturation of pests to some extents. Take oriental fruit fly as an example; rainy days lead to their maturation. In contrast, low temperature results in a small group of oriental fruit fly. As a whole, in building this model, we presumed that there exists the connection between weather and group size of the pest.
  • Features:
          The features we chose to utilize included average day temperature, the highest day temperature, the lowest day temperature, rainfall, pressure, wind velocity and relative humidity.
  • Model selection:
          We are using two famous and frequently used models, Supported Vector Machine (SVM) and Random Forest (RF) in building the prediction system. In processing and testing data, the member of NYMU discovered that SVM made an incredibly great prediction (~100%) when the datasets are big enough, while RF has an average score of 80%. However, in some cases, when RF model reached 60~70% of accuracy score, SVM has 50% or worse. Thus, we decided to build a simple ensemble prediction model, with different hyperparameters and classifier. Moreover, choose the pest population size that most classifiers agree.

    • Distribution Kit

    NTHU_Taiwan iGEM team

    We provided NTHU_Taiwan iGEM team with BBa_B0012.
    (NTHU linker)

    Mingdao

    We sent the biobricks BBa_K1694002, BBa_K1694010, BBa_K1694035, BBa_K1694044 and BBa_K1694045, which was registry by 2015 NCTU_Formosa, to help them express the outer membrane protein.
    (Mingdao linker)

    NCKU_Tainan

    This year, it is the first time for NCKU_Tainan to participate in Giant Jamboree. Therefore, we provided them with the Part BBa_K861171, BBa_I712667, BBa_I712015, BBa_R0010, BBa_R0040, BBa_C0040, BBa_C0051, BBa_C0061, BBa_C0062, BBa_R0065, BBa_K592009 and BBa_R0052 to them.
    (NCKU_Tainan linker)

    Mentoring a New Team