The invention belongs to the technical field of sandstone type
uranium ore mineralization prediction, and particularly relates to a sample set construction method suitable for sandstone type
uranium ore
big data prediction, which comprises the following steps: a
system collects and arranges geological-
geophysical prospecting-geochemical
prospecting-
remote sensing data of a research area, extracts the geological,
geophysical prospecting, geochemical
prospecting and
remote sensing data, and pre-processes the data; performing
feature selection of a sample set by analyzing factors such as construction,
magma activity, stratum and lithofacies paleogeography,
geophysics,
geochemistry and the like; integrating the selected sample
feature data, and marking features and labels of each sample to form complete
single sample data; and integrating the formed
single sample data together to construct a sample set, performing
feature engineering on the data in the sample set, randomly selecting samples to construct a
training set, and using the remaining samples to construct a
test set. According to the method, the working process of constructing the sandstone type
uranium mine
big data prediction sample set is optimized, so that the uncertainty in the working process is reduced.