A. Relationship between crowdsourced and AI-imputed maize prices, disaggregated by state and commodity subtype within the northern region of Nigeria.

Crowdsourced prices, submitted daily by volunteers over a 3-year period (2019–2021), are averaged into monthly values. High correlations were observed across all states and by commodity subtypes. The correlation coefficient (r ranging from -1 to + 1) and the coefficient of determination r² (ranging from 0-1) quantify the reliability of AI-estimated prices. The straight lines represent best linear line of fit between the correlated price data pairs. The p-value indicates the significance of the relationship at α =  0.05. A similar figure is available for rice in the supporting information (S2 Fig). B. Map of study area, with Admin 2 boundaries, in northern Nigeria, showing (i) local correlation between crowdsourced and AI-imputed maize prices at the nearest market, and (ii) log-transformed count of submissions in the crowdsourced dataset, indicating the sample size underpinning the average maize prices in the sub-national correlation analysis. The value for counts were log-transformed to normalize the skewed distribution and improve visual representation of the map. Both maps show that the target market locations for AI-imputation are within Admin 2 boundaries where there is moderate to high count of submissions and where high correlations were achieved. The correlation coefficient (r, ranging from -1 to + 1) was calculated based on monthly average prices from both data sources over a 36-month period (January 2019–December 2021). The administrative boundary outlines are mapped using shapefile data from GADM, available under a free license (www.gadm.org). C. Spatial distribution of crowdsourced maize prices across three major market segments within the northern region of Nigeria. Each point represents a georeferenced price quote submitted by volunteers through a smartphone-based app. The administrative boundary outlines are mapped using shapefile data from GADM, available under a free license (www.gadm.org). D. Relationship between crowdsourced and AI-imputed maize prices across three major market segments within the northern region of Nigeria. Crowdsourced prices were collected from volunteer submissions over a 3-year period (2019–2021) and averaged into monthly values. The AI-imputed monthly close prices were averaged over four market locations within the study region. N_Cr denotes the number of crowdsourced data points included in the analysis during the period. The coefficient of determination (r²; ranging from 0-1) was tested for significance at α =  0.05. The straight lines represent best line of fit between the correlated price data pairs, while the grey area around the line represent 95% confidence interval for each point on the regression line. The figure shows strong covariance between both price datasets in each of the market segments.