multiple-regression(): first arg is now a list of N-tuples (each N-tuple

representing one input (setting of indep vars to values). The returned
predictor fn also takes an N-tuple brownplt#1732

src/js/trove/multiple-regression.js

@@ -132,8 +132,8 @@
       let Y = new Array(num_mappings);
       let x_s_len = false;
       js_x_s_s.forEach(function(x_s, i) {
-        runtime.checkList(x_s);
-        let js_x_s = runtime.ffi.toArray(x_s);
+        runtime.checkTuple(x_s);
+        let js_x_s = x_s.vals;
         let x_s_n = js_x_s.length;
         if (x_s_len === false) {
           x_s_len = x_s_n;
@@ -156,8 +156,8 @@
       let XT = matrixTranspose(X);
       let B = matrixMultiply2(matrixInverse(matrixMultiply2(XT, X)), matrixMultiply2(XT, Y));
       let Bfunc = function(x_s) {
-        runtime.checkList(x_s);
-        let js_x_s = runtime.ffi.toArray(x_s);
+        runtime.checkTuple(x_s);
+        let js_x_s = x_s.vals;
         if (js_x_s.length !== x_s_len) {
           throw runtime.ffi.throwMessageException("predictor: wrong number of arguments");
         }

tests/pyret/tests/test-statistics.arr

@@ -114,10 +114,10 @@ check "linear regression":
 end
 
 check "multiple regression":
-  x-s-s = [list: [list: 4], [list: 4.5], [list: 5], [list: 5.5], [list: 6], [list: 6.5], [list: 7]]
+  x-s-s = [list: {4}, {4.5}, {5}, {5.5}, {6}, {6.5}, {7}]
   y-s   = [list: 33, 42, 45, 51, 53, 61, 62]
   pf = multiple-regression(x-s-s, y-s)
-  pf([list: 8]) is%(within-abs(0.001)) 73.3214
+  pf({8}) is%(within-abs(0.001)) 73.3214
 end