Pest Population Data for Machine Learning Benchmarking in Turkish Agriculture, 2020-2022

An agricultural dataset containing meteorological and vegetation variables was used as a representative case study for comparing AI-based machine learning methods. The data is based on population observations of Cimbex quadrimaculata in Diyarbakır and Elazığ provinces in Türkiye between 2020 and 2022. Three different modeling approaches (binary classification, multiclass classification, and regression) were applied to the same data to enable systematic comparison of model performance, generalizability, and explainability.

Use Cases

• Benchmarking Gradient Boosting, XGBoost, and LightGBM performance based on the described binary classification, multiclass classification, and regression tasks.
• Applying SHAP analysis for model interpretability based on the identified influential temperature- and humidity-related variables.
• Studying the capture of nonlinear relationships and nested effects in ecological data based on the described complex data structures.
• Comparing model evaluation metrics (accuracy, F1, AUC, R², RMSE, MAE) under cross-validation frameworks as detailed in the study.

Strengths

• Dataset is explicitly designed for a three-approach ML comparison (binary classification, multiclass classification, regression) on the same data.
• Specific model performance metrics are reported, including Gradient Boosting achieving 94.3% accuracy and 0.983 AUC in binary classification.
• SHAP analysis was conducted, identifying temperature- and humidity-related variables as consistently influential predictors.
• Data has a clear temporal (2020-2022) and geographic (Diyarbakır and Elazığ, Türkiye) scope.

Limitations

• Row count and column-level documentation are unknown, which limits suitability assessment and requires manual inspection after download.
• The dataset is very small at 5.5 KB, indicating limited scope and sample size.
• Column names and sample data are unavailable, so field semantics must be inferred from the description or the downloaded file.

Provenance

Source

Author Yunus Güral, published on figshare.

Collection Method

Population observations of Cimbex quadrimaculata (a pest species) combined with meteorological and vegetation variables.

Time Range

2020 to 2022

Freshness

Last updated 2026-04-03 17:48:58; freshness should be verified.

Geography

Diyarbakır (Eğil) and Elazığ (Keban) provinces, Türkiye