Developing V0 of the Decision Support Tools.
Download as PPT, PDF1 like81 views
This document provides an overview and status update of efforts to develop Version 0 (V0) of Decision Support Tools (DST) for cassava. It describes activities including a literature review that collected over 1,200 papers on cassava agronomy, development of software architecture documents to guide tool development, collection of 224 farmer scenarios, and initial meta-analyses to derive relationships for predicting outcomes. Next steps include continuing the literature review and meta-analysis, handing results to developers, publishing review reports, and extracting lessons to improve further development of V1 tools.
Developing V0 of the Decision Support Tools.
- 1. www.iita.orgA member of CGIAR consortium Developing V0 of the Decision Support Tools Current progress and planned activities Pieter Pypers, 06-12-2016
- 2. www.iita.orgA member of CGIAR consortium V0 DST concept Literature review Meta-analysis What is V0? • Repository of relevant cassava agronomy literature on DMS accessible to all members • Extraction of relevant information into a database (continuously updated) • Meta-analysis to derive general relationships relevant to use cases I = f0(E,…) • Concept of a decision support system (not a tangible tool) • Functionality and requirements defined together with development partners (co-development) • Predictions based on meta-analysis on studies from literature
- 3. www.iita.orgA member of CGIAR consortium Co-development process Scenario descriptionsScenario descriptions Context description Question? Format of support Relevant information Dev partner leaders Dev partner leaders Extension agents Extension agents Cassava growers Cassava growers Software developers Software developers ResearchersResearchers GeneralizationGeneralization Context & GIS input Request specs Output Client input pars • Interactions with project clients (dev partners) • Stakeholder meeting • Survey to collect scenarios • Summarize scenarios • Feedback sessions • Define functional requirements • Software architecture docs
- 4. www.iita.orgA member of CGIAR consortium Co-development process
- 5. www.iita.orgA member of CGIAR consortium Co-development process 224 scenarios collected across use cases224 scenarios collected across use cases Scenarios collected: survey.reevehost.com (open registration and access)
- 6. www.iita.orgA member of CGIAR consortium Co-development process Survey summary results
- 7. www.iita.orgA member of CGIAR consortium Tool development Software architecture documents developed for each use case tool These documents describe all the details on functional requirements to allow software developers to start implementing the decision support tools.
- 8. www.iita.orgA member of CGIAR consortium Tool development Software architecture example: Scheduled Planting & High Starch Log in / registration Capture location Define request Monitor usage Capture client inputRetrieve GIS layers Process Provide DS output Collect feedback
- 9. www.iita.orgA member of CGIAR consortium Tool development V0 fertilizer recommendation tool based on QUEFTS modelling Currently, site-specific recommendations are being evaluated in validation trials in Tanzania. 1 2 3 4 5
- 10. www.iita.orgA member of CGIAR consortium Literature review 1,267 reports and papers collected on cassava agronomy, physiology and research methodology 1,267 reports and papers collected on cassava agronomy, physiology and research methodology
- 11. www.iita.orgA member of CGIAR consortium Literature review Data source When and where conducted? Variety characteristics, agronomic parameters Root yield (starch content) + measure of precision Data extracted from literature (continuous exercise) Nr studies digitized relevant to ACAI use cases: FR: 47 IC: 35 SP: 26 FB: 0 PP: 31 HS: 30 Countries: Benin 2 Brazil 2 China 9 Colombia 10 Costa Rica 1 DRC 1 Ethiopia 1 Ghana 17 India 10 Indonesia 7 Kenya 3 Malaysia 20 Mozambique 2 Nigeria 29 Philippines 7 Sierra Leone 2 Sri Lanka 1 Tanzania 2 Thailand 19 Togo 4 Trinidad and Tobago 1 Uganda 3 Vietnam 14 Zambia 1 Nr studies digitized relevant to ACAI use cases: FR: 47 IC: 35 SP: 26 FB: 0 PP: 31 HS: 30 Countries: Benin 2 Brazil 2 China 9 Colombia 10 Costa Rica 1 DRC 1 Ethiopia 1 Ghana 17 India 10 Indonesia 7 Kenya 3 Malaysia 20 Mozambique 2 Nigeria 29 Philippines 7 Sierra Leone 2 Sri Lanka 1 Tanzania 2 Thailand 19 Togo 4 Trinidad and Tobago 1 Uganda 3 Vietnam 14 Zambia 1
- 12. www.iita.orgA member of CGIAR consortium Meta-analysis Example: High starch content 449 data points from 28 studies on 173 varieties Predict starch yield based on root yield: StarchY = 0.285*rootY – 0.000498*rootY2 + σvariety + σenvironment More variation due to environment (σenvironment = 1.48 t/ha) than due to variety (σvariety = 0.57 t/ha) Fertilizer N application tends (P = 0.12) to negatively affect starch contents but effect is smaller (-380 kg/ha starch per 100 kg/ha fertilizer N) than response to N (910 kg/ha starch per 100kg/ha fertilizer N). 106 data points from 10 studies on 22 varieties with varying N levels applied
- 13. www.iita.orgA member of CGIAR consortium Planned activities • Continue digitization of results and update/improve meta-analysis • Hand over meta-analysis results (regression trees or prediction models) to software developers as a starting point to develop V1 • Prepare a review report document (basis = literature review by PhD students on individual use cases) • Publish paper on learnings from meta-analysis • Extract lessons to improve/direct ongoing activities (continuous) What’s next?
- 14. www.iita.orgA member of CGIAR consortium