/
Plant Prodigies

Plant Prodigies

 

The Ten Rules

  • Rule 1: Develop reflexive habits

    • We felt that we did not effectively follow this rule. Collaboration between teams felt “stiff” as to give the impression that every other team knew what they were already doing. Although there were ample resources available online and through our slack channel, we often found ourselves asking ‘what to ask’ when looking for help.

  • Rule 2: Communicate the project management plan early and often

    • We did not have defined and reachable goals from the beginning, therefore this caused the group to underperform throughout the assignment. We heavily relied on an LSTM model built for stocks but should have sought other models that deal with plant science and crops. For future assignments, all group members should create Standard Operating Procedures (SOPs)–which should help strengthen collaboration, feedback, and lessen bottlenecks.

  • Rule 3: Speak the same language

    • We believe the atmosphere of the team was open to all ideas and questions. Some of us were a bit more experience with ML, therefore, those team members helped explain the jargon of ML to those with little to no experience with ML. We were open via two channels of communication Slack and Discord, therefore the team did not have issues with communication. Everyone’s opinion was respected and if any conflicts came up–they were quickly resolved to avoid hindrance to the team’s progress.

  • Rule 4: Design the project so that everybody benefits

    • Knowing that we all had little to no experience with ML, this project was a difficult one since we are all inexperienced. We tried our best to not let this impact our collaboration or contributions to the midterm. Throughout the project we all experienced frustration but knowing that complaining won’t help, we decided to reach out to other teams to get some assistance. Regrettably, we should have also reached out to our TA initially, since the team would’ve benefited tremendously to avoid all the bottlenecks we went through. In hindsight, this project ultimately taught us the fundamentals of creating a ML model using cyberinfrastructure applications.

  • Rule 5: Fail early and often

    • Although we did encounter failures, we felt that we did not encounter them often enough. Most of our failures involved problems with the paper’s LSTM model and attempting to replicate that. These failures are a major contributor on most of the early slow progress. Had we encountered these failures earlier enough, perhaps we wouldn’t waste so much time on this bottleneck.

  • Rule 6: Share collaboration tools

    • We definitely followed this rule closely. Although it wasn’t very meaningful among other teams, our team made a lot of research into ML concepts and collaborated within our own team. A lot of progress was made by stacking each others research. Getting the team up to date with the technology stack used was also important during this project.

  • Rule 7: Manage your data like the collaboration depends on it

    • We are not sure that we followed this rule. At the beginning, we felt that understanding the data was important, as a lot of our early research went into what each training and test data given meant. As other have encountered problems with the data, we attempted to pivot to trying to understand other models to use. When we returned to the data, we realized that organizing and cleaning the data was just as important. In retrospect, an equal priority should be given to understanding the data and the models to use in our project.

  • Rule 8: Write code that others can use and reproduce

    • The reproducibility and reusability of code is important, but we felt that as a team this rule was put to low priority. Most of the code we’ve written was focused on getting the necessary functions to perform. As a result, a majority of our code went messy and unorganized. There was a good effort to clean up and organize our code at the end.

  • Rule 9: Observe ethical hygiene

    • As a team, we felt that maintaining good ethics was an important factor during our work on this project. Although the project data did not really apply to this rule, we followed our work patterns along this line. This mostly involved making good efforts to schedule out time to research and collaborate, at the same time respecting the scheduling difficulties of others.

  • Rule 10: Document your collaboration

    • We’re not sure if we followed this rule correctly. We felt that most of our documentation was either through formal documents (the README, Midterm write up), or among our peers through instant messaging group chat. In retrospect, made more meaningful comments in our code could have been made to follow our general thought/progress process, as well as for reproducibility.

 

  • Strengths:

    • Data manipulation and analysis

    • Python programming

    • Database creation and management

  • Possible Weaknesses:

    • AI and Machine Learning

    • Expertise of plant sciences

    • Linux