<result> <column_name>SP</column_name> <family>uncertain</family> <subtype>none</subtype> <confidence>0.35</confidence> <requires_user_confirmation>true</requires_user_confirmation> <sum_meaningful>false</sum_meaningful> <signed>false</signed> <reducers> <primary_reducer>mean</primary_reducer> <valid_reducers>mean, min, max, median, distribution</valid_reducers> <excluded_reducers>product (not a ratio; values far from 1.0), geomean (not a ratio), sum (unconfirmed; may be a level rather than an accumulating quantity)</excluded_reducers> <average_semantics>arithmetic_mean</average_semantics> </reducers> <inner_outer_measure> <inner_measure_reducer>min</inner_measure_reducer> <outer_measure_reducer>max</outer_measure_reducer> </inner_outer_measure> <multiplicative_partner_derivable>false</multiplicative_partner_derivable> <reasoning> The table is PATH with a Date sequence axis, so the full analysis applies, but there are no reconstruction statistics and no unit. The name "SP" is an opaque abbreviation and gives no naming signal. Multiplicative is ruled out: mean 74 and median 70 are far from 1.0, the range runs to 253, there are zeros, and log_symmetry_gain is low. The values are non-negative, integer-like and bounded, which fits an absolute level (intensive_additive, or ordinal_additive on a PATH table) or a count-like quantity (real_additive). Because the data cannot separate these, I abstain on the subtype and use conservative level-style reducers pending human confirmation of what SP measures. </reasoning> <table_type_conflict>false</table_type_conflict> </result>