Experimentenbibliotheek
Deze pagina bundelt alle experiments en hypotheses uit de repository. Elk kaartje verwijst naar de ruwe markdown en een eventuele Codex-kwalificatie.
Experiments
FAMILY_MULTI_SPECTRAL_INDICES
**Hypothesis**: Multiple vegetation and moisture indices (NDVI, EVI, NDWI, SAVI, GNDVI, NBR) from Sentinel-2 provide stronger predictive signal than single NDVI by capturing different crop stress dimensions (water, chlorophyll, soil moisture), achieving 10-15% improvement over persistent baseline at 12-week horizon.
Experiments
FAMILY_SATELLITE_ANOMALY_DETECTION
**Core Hypothesis**: While normal vegetation variation does not predict continuous potato prices (as proven by NDVI and multi-spectral failures), **extreme satellite-visible anomalies** that represent catastrophic events (droughts, floods, frost) do precede major price spikes. By focusing on binary classification of extreme events rather than continuous regression, satellite data can provide early warning of price disruptions.
Experiments
FAMILY_SATELLITE_STORAGE_INFERENCE
**CRITICAL**: This experiment will use REAL DATA ONLY from repository interfaces. NO synthetic/mock data allowed.
Experiments
FAMILY_SEASONAL_PLANTING
This experiment tests whether multi-temporal vegetation indices (NDVI, EVI, SAVI, GNDVI) measured during potato growth and harvest periods (May-October) contain predictive signals for potato prices 1-3 months into the storage season (December-February). Based on literature findings showing R² = 0.67-0.84 between peak NDVI and yield, combined with established yield-price relationships, we hypothesize that growth season satellite monitoring provides valuable forward-looking price signals with 5-15% MAPE improvement potential.
Experiments
FAMILY_STORAGE_DECAY_INTEGRATION
**CRITICAL**: This experiment uses ONLY REAL DATA from repository interfaces. NO synthetic/mock data allowed.
Experiments
FAMILY_YOY_NDVI_LANDSAT
**Hypothesis**: Integrating Landsat NDVI intelligence with the proven YoY methodology will achieve **57-62% improvement** by capturing multi-decadal vegetation patterns that drive year-over-year potato price changes. Building on the revolutionary 54.6% weather success, NDVI satellite intelligence should provide an additional 3-8 percentage points through 40-year historical vegetation analysis combined with proven YoY framework.
Experiments
FAMILY_YOY_PRICE_ONLY
**Hypothesis**: Year-over-Year (YoY) price change targets using the formula `Y_{t,m+h} = 100 × (P_{t,m+h} - P_{t-1,m+h}) / P_{t-1,m+h}` create superior predictive power by focusing on relative annual patterns rather than absolute prices, leveraging seasonal cycles and storage dynamics that repeat yearly. This approach should achieve 10-30% improvement over standard baselines at 8-week horizons.
Experiments
FAMILY_YOY_WEATHER
**Hypothesis**: Adding weather intelligence to the proven YoY price framework will achieve **>35% improvement** by capturing year-over-year climate variations that drive potato price changes. Weather-enhanced YoY targets should outperform the revolutionary 32.8% success of FAMILY_YOY_PRICE_ONLY by 3-13 percentage points through agricultural weather pattern recognition.