agrodesign

AgroDesign

AgroDesign is a unified analysis framework for agricultural experiments.

Instead of manually choosing statistical tests, you describe the experiment — AgroDesign automatically performs the correct analysis and produces biological recommendations.

It converts experimental data directly into decisions, reports, figures, and publication-ready outputs.

---

Install

pip install agrodesign

---

5-Minute Example

Randomized Complete Block Design (variety trial)

import pandas as pd
from agrodesign.experiment import Experiment

df = pd.DataFrame({
    "Block":["B1","B1","B1","B1",
             "B2","B2","B2","B2",
             "B3","B3","B3","B3",
             "B4","B4","B4","B4"],
    "Variety":["V1","V2","V3","V4"]*4,
    "Yield":[48,52,56,60,45,50,54,58,47,51,55,59,46,49,53,57]
})

result = Experiment(df,"Yield").rcbd("Variety","Block").run()

result

Output:

AgroResult (RCBD)
Response: Yield
Significant factors: Variety
Best treatment: V4
Expected yield: 58.50

---

What can you do next?

print(result)      # Full scientific report
result.summary()   # Agronomic recommendation
result.plot()      # Publication figures
result.export("report")  # Tables + plots folder

---

Universal Workflow

All analyses follow the same pattern:

result = (
    Experiment(data, response)
    .design(...)
    .run()
)

---

Supported Experimental Designs

---

Example Analyses

Factorial experiment

Experiment(df,"Yield").factorial(["Nitrogen","Spacing"]).run()

Mixed model (adjusted performance)

Experiment(df,"Yield").mixed(fixed=["Treatment"], random=["Block"]).run()

Multi-environment trial (stability)

Experiment(df,"Yield").gxe("Genotype","Environment","Rep").run()

Multi-year experiment

Experiment(df,"Yield").by("Year").rcbd("Variety","Block").run()

Multi-trait selection

Experiment(df,["Yield","Height"]).rcbd("Variety","Block").run()

---

What AgroDesign Handles Automatically

• Correct ANOVA model construction
• Error-term selection
• Mean separation (LSD, Tukey, DMRT)
• Interaction interpretation rules
• Mixed-model BLUP ranking
• G×E stability analysis (AMMI, GGE, FW, ER)
• Assumption diagnostics
• Publication-ready plots

No manual statistical decisions required.

---

Philosophy

Traditional workflow:

Choose model → run statistics → interpret biology

AgroDesign workflow:

Describe experiment → AgroDesign chooses statistics → interpret biology

---

Citation

If you use AgroDesign in academic work:

AgroDesign v0.6.0 — Stable Research Release https://github.com/DeepStatistix/AgroDesign

(DOI will be added via Zenodo)

---

License

MIT License

---

Author

Aqib Gul DeepStatistix