Data commit

Task/Merge-and-aggregate-datasets/Datalog/merge-and-aggregate-datasets-1.datalog

@@ -0,0 +1,10 @@
+// datetime.cpp
+#include <ctime>
+#include <cstdint>
+extern "C" {
+  int64_t from date(const char* string) {
+    struct tm tmInfo = {0};
+    strptime(string, "%Y-%m-%d", &tmInfo);
+    return mktime(&tmInfo); // localtime
+  }
+}

Task/Merge-and-aggregate-datasets/Datalog/merge-and-aggregate-datasets-2.datalog

@@ -0,0 +1,36 @@
+/* rosetta.dl */
+#define NaN to_float("NaN")
+.functor from_date(date:symbol) : number
+.decl Patient(id:number, lastname:symbol)
+.decl Visit(id:number, date:symbol, score:float)
+.decl SummaryDates(id:number, lastname:symbol, last_date:symbol)
+.decl SummaryScores(id:number, lastname:symbol, score_sum:float, score_mean:float)
+.decl MissingDates(x:number)
+
+Patient(1001,"Hopper").
+Patient(4004,"Wirth").
+Patient(3003,"Kemeny").
+Patient(2002,"Gosling").
+Patient(5005,"Kurtz").
+
+Visit(2002,"2020-09-10",6.8).
+Visit(1001,"2020-09-17",5.5).
+Visit(4004,"2020-09-24",8.4).
+Visit(2002,"2020-10-08",NaN).
+Visit(1001,"",6.6).
+Visit(3003,"2020-11-12",NaN).
+Visit(4004,"2020-11-05",7.0).
+Visit(1001,"2020-11-19",5.3).
+
+MissingDates(@from_date("")) :- true.
+
+SummaryDates(id, lastname, last_date) :-
+     Patient(id,lastname),
+     last_timestamp = max ts: {Visit(id, date, _), ts = @from_date(date), !MissingDates(ts)},
+     Visit(id, last_date, _), last_timestamp = @from_date(last_date).
+SummaryScores(id, lastname, score_sum, score_mean) :-
+     Patient(id,lastname),
+     score_sum = sum score: {Visit(id, _, score), score != NaN},
+     score_mean = mean score: {Visit(id, _, score), score != NaN}.
+.output SummaryDates
+.output SummaryScores

Task/Merge-and-aggregate-datasets/Datalog/merge-and-aggregate-datasets-3.datalog

@@ -0,0 +1,2 @@
+g++ -shared -fPIC datetime.cpp -o libfunctors.so
+souffle -D- rosetta.dl