Click on the link below to view the webinar by Lily Clements from IDEMS International on 25 September 2025 entitled Demonstrating Tricot Analysis in R-Instat.
Presenter Bio: Lily is a Data Scientist and Educator at IDEMS. She works at the intersection of data tools, education, and social impact – creating open-source software and accessible resources that make data skills easier to understand, teach, and use responsibly in the real world.
For more info, contact Lily Clements at lily@idems.international
Summary of presentation: Lily demonstrated how to use R-Instat, a user-friendly R-based statistical software, for Tricot data analysis using real agricultural data (cassava trials in Nigeria and Cameroon).
Useful Links:
Download R-instat: https://r-instat.org/
Lily’s presentation: 1000FARMS 2025 Webinar Series template.pptx - Google Slides
| Question | Answer |
|---|---|
| I used ClimMob to do randomization. We don’t rank best and worst but a 1 to 5 score for a package. Could you transform rankings of 5 instead of 3 in R-instat? | Yes, Roger Stern (IDEMS) tested it by using 4 rankings and it works well with multiple rankings. |
| Did we need to write some script to use R-instat? | No, you don’t need to do any coding yourself. You can install it from online and it bundles with the R version so you can just go from there. However, if you want to see the R code there is a dialog window that allows that. You can therefore modify the code or just see what is being run. |
| What is new about R-instat if you are a programmer and already familiar with R? | I also use R but still find R-instat very useful for things such as the Prepare menu and it actually has a very extensive menu when it comes to graphs and tables for creating your plots and for customizing. Given there are more than 20,000 packages on Cran now, it is sometimes hard to find the best package to do things and R-instat can help with that. Also you can suggest other things you’d want to see in it. |
| Is this only for tricot analysis, or it has another capabilities, such as multi environment trial analysis, single trial analysis with more genotypes, or is it in the plan to include this in the tool? | Basically I can do my general analysis and modeling in here—just the usual stuff. But if my data has a specific structure, like tricot data, then there are dedicated menus for that. So there’s the tricot menu, there’s a climatic menu for daily rainfall, temperature, that sort of thing. The idea is that it would make sense to have a new menu for multi-environment trial analyses— that doesn’t exist yet, but it could be designed, just like the tricot menu was. The software is set up for these tailored menus that match a specific data format and make the right analyses easier. Also, tricot is very specific to ranking data; if I’ve got rating data, that’s a different nature of problem. And there are a couple of Shiny apps out there that cover bits of multi-environment analysis—would be great to fold them into one place, especially for people who don’t code, and bring in previous analyses (like genotype choices) alongside the tricot work. |
| How does R-instat actually track data - is the metadata coming with the data? | David Stern (IDEMS): The key difference from working in plain R is that the system uses metadata to link multiple data frames so they behave like one—more like a small database than a single table—which lets it choose and run the right analyses based on how the pieces relate. If I import data straight from CLimMob, the known metadata auto-defines the structures, so I don’t have to do manual transformations. A big part of making analysis easier is keeping the data organized by level—individual/farm, plot (where varieties differ within farms), and variety—because Tricot workflows rely on all three. Much of the complexity comes from juggling those levels and using each appropriately. |
| Can the platform handle quantitative traits like yield—in addition to the current qualitative ranking traits? | If its something you can do in R then you should be able to do it in R-instat. I’ve not tried that specific case, so I’d love to know how it goes, or if you do it and and if it doesn’t work, then it would be great for you to let us know on GitHub so that we can fix that. |
| How flexible would it be to incorporate new functionalities in the future in this platform? | The platform is open source and flexible, so anyone can contribute, but creating custom dialogs/menus is currently a high barrier. The team plans to make this easier—potentially with AI agents to help maintain and upgrade tailored menus—but that’s still in development. For now, it operates like a standard OSS project: contributors can learn the workflow, and support is available from the team, including trained groups in Kenya and Ghana. |
| Are we able to add extra columns (e.g., yield or other variables) directly in the interface’s table-like view, so additional data can be included for analysis? | Yes—you can add extra columns. Under Prepare (Data Frame) you can insert new columns, or use the Calculator; you can also merge in external data. The tricot data structure (farm, plot, variety) actually makes it easier to add data at the right level—plot-, farm-, or variety-level—while staying within tricot functionality. Instat itself is general-purpose and handles large datasets; practical limits are essentially those of R, not the interface. |