@Test
public void testCatGroup() {
Frame fr = null;
String tree =
"(GB hex [4] nrow 0 \"all\" mean 2 \"all\")"; // Group-By on col 4, no order-by, nrow and
// mean of col 2
try {
fr = chkTree(tree, "smalldata/iris/iris_wheader.csv");
chkDim(fr, 3, 3);
chkFr(fr, 0, 0, "Iris-setosa");
chkFr(fr, 1, 0, 50);
chkFr(fr, 2, 0, 1.464);
chkFr(fr, 0, 1, "Iris-versicolor");
chkFr(fr, 1, 1, 50);
chkFr(fr, 2, 1, 4.26);
chkFr(fr, 0, 2, "Iris-virginica");
chkFr(fr, 1, 2, 50);
chkFr(fr, 2, 2, 5.552);
fr.delete();
fr = chkTree("(GB hex [1] mode 4 \"all\" )", "smalldata/iris/iris_wheader.csv");
chkDim(fr, 2, 23);
} finally {
if (fr != null) fr.delete();
Keyed.remove(Key.make("hex"));
}
}
static Frame exec_str(String str, String id) {
Val val = Exec.exec(str);
switch (val.type()) {
case Val.FRM:
Frame fr = val.getFrame();
Key k = Key.make(id);
// Smart delete any prior top-level result
Iced i = DKV.getGet(k);
if (i instanceof Lockable) ((Lockable) i).delete();
else if (i instanceof Keyed) ((Keyed) i).remove();
else if (i != null)
throw new IllegalArgumentException("Attempting to overright an unexpected key");
DKV.put(fr = new Frame(k, fr._names, fr.vecs()));
System.out.println(fr);
checkSaneFrame();
return fr;
case Val.NUM:
System.out.println("num= " + val.getNum());
assert id == null;
checkSaneFrame();
return null;
case Val.STR:
System.out.println("str= " + val.getStr());
assert id == null;
checkSaneFrame();
return null;
default:
throw water.H2O.fail();
}
}
@Test
public void testBasicDdply() {
Frame fr = null;
String tree =
"(ddply hex [1] { x . (mean (cols x 2) TRUE)})"; // Group-By on col 1 (not 0) mean of col 2
try {
fr = chkTree(tree, "smalldata/iris/iris_wheader.csv");
chkDim(fr, 2, 23);
chkFr(fr, 0, 0, 2.0); // Group 2.0, mean is 3.5
chkFr(fr, 1, 0, 3.5);
chkFr(fr, 0, 1, 2.2); // Group 2.2, mean is 4.5
chkFr(fr, 1, 1, 4.5);
chkFr(fr, 0, 7, 2.8); // Group 2.8, mean is 5.043, largest group
chkFr(fr, 1, 7, 5.042857142857143);
chkFr(fr, 0, 22, 4.4); // Group 4.4, mean is 1.5, last group
chkFr(fr, 1, 22, 1.5);
fr.delete();
fr =
chkTree(
"(ddply hex [1] { x . (sum (* (cols x 2) (cols x 3)))})",
"smalldata/iris/iris_wheader.csv");
chkDim(fr, 2, 23);
} finally {
if (fr != null) fr.delete();
Keyed.remove(Key.make("hex"));
}
}
@Test
public void testImpute() {
Frame fr = null;
try {
// Impute fuel economy via the "mean" method, no.
String tree = "(h2o.impute hex 1 \"mean\" \"low\" [])";
fr = chkTree(tree, "smalldata/junit/cars.csv");
chkDim(fr, 8, 406);
Assert.assertEquals(0, fr.vec(1).naCnt()); // No NAs anymore
Assert.assertEquals(23.51, fr.vec(1).at(26), 1e-1); // Row 26 was an NA, now as mean economy
fr.delete();
// Impute fuel economy via the "mean" method, after grouping by year. Update in place.
tree = "(h2o.impute hex 1 \"mean\" \"low\" [7])";
fr = chkTree(tree, "smalldata/junit/cars.csv");
chkDim(fr, 8, 406);
Assert.assertEquals(0, fr.vec(1).naCnt()); // No NAs anymore
Assert.assertEquals(
17.69, fr.vec(1).at(26), 1e-1); // Row 26 was an NA, now as 1970 mean economy
} finally {
if (fr != null) fr.delete();
Keyed.remove(Key.make("hex"));
}
}
@Test
public void testAllAggs() {
Frame fr = null;
try {
String tree =
"(GB hex [4] nrow 0 \"rm\" mean 1 \"rm\" sum 1 \"rm\" min 1 \"rm\" max 1 \"rm\" )";
fr = chkTree(tree, "smalldata/iris/iris_wheader.csv");
chkDim(fr, 6, 3);
chkFr(fr, 0, 0, "Iris-setosa");
chkFr(fr, 1, 0, 50); // nrow
chkFr(fr, 2, 0, 3.418); // mean
chkFr(fr, 3, 0, 170.9); // sum
chkFr(fr, 4, 0, 2.3); // min
chkFr(fr, 5, 0, 4.4); // max
chkFr(fr, 0, 1, "Iris-versicolor");
chkFr(fr, 1, 1, 50); // nrow
chkFr(fr, 2, 1, 2.770); // mean
chkFr(fr, 3, 1, 138.5); // sum
chkFr(fr, 4, 1, 2.0); // min
chkFr(fr, 5, 1, 3.4); // max
chkFr(fr, 0, 2, "Iris-virginica");
chkFr(fr, 1, 2, 50); // nrow
chkFr(fr, 2, 2, 2.974); // mean
chkFr(fr, 3, 2, 148.7); // sum
chkFr(fr, 4, 2, 2.2); // min
chkFr(fr, 5, 2, 3.8); // max
} finally {
if (fr != null) fr.delete();
Keyed.remove(Key.make("hex"));
}
}
@Test
public void testNAHandle() {
Frame fr = null;
try {
String tree =
"(GB hex [7] nrow 0 \"all\" mean 1 \"all\")"; // Group-By on year, no order-by, mean of
// economy
fr = chkTree(tree, "smalldata/junit/cars.csv");
chkDim(fr, 3, 13);
chkFr(fr, 0, 0, 70); // 1970, 35 cars, NA in economy
chkFr(fr, 1, 0, 35);
chkFr(fr, 2, 0, Double.NaN);
chkFr(fr, 0, 2, 72); // 1972, 28 cars, 18.714 in economy
chkFr(fr, 1, 2, 28);
chkFr(fr, 2, 2, 18.714, 1e-1);
fr.delete();
tree = "(GB hex [7] nrow 1 \"all\" nrow 1 \"rm\" nrow 1 \"ignore\")"; // Group-By on year, no
// order-by, nrow of
// economy
fr = chkTree(tree, "smalldata/junit/cars.csv");
chkDim(fr, 4, 13);
chkFr(fr, 0, 0, 70); // 1970, 35 cars, 29 have economy
chkFr(fr, 1, 0, 35); // ALL
chkFr(fr, 2, 0, 29); // RM
chkFr(fr, 3, 0, 29); // IGNORE
fr.delete();
tree = "(GB hex [7] mean 1 \"all\" mean 1 \"rm\" mean 1 \"ignore\")"; // Group-By on year, no
// order-by, mean of
// economy
fr = chkTree(tree, "smalldata/junit/cars.csv");
chkDim(fr, 4, 13);
chkFr(fr, 0, 0, 70); // 1970, 35 cars, 29 have economy
chkFr(fr, 1, 0, Double.NaN); // ALL
chkFr(fr, 2, 0, 17.69, 1e-1); // RM
chkFr(fr, 3, 0, 14.66, 1e-1); // IGNORE
} finally {
if (fr != null) fr.delete();
Keyed.remove(Key.make("hex"));
}
}
@Test
public void testBasic() {
Frame fr = null;
String tree =
"(GB hex [1] mean 2 \"all\")"; // Group-By on col 1 (not 0), no order-by, mean of col 2
try {
fr = chkTree(tree, "smalldata/iris/iris_wheader.csv");
chkDim(fr, 2, 23);
chkFr(fr, 0, 0, 2.0); // Group 2.0, mean is 3.5
chkFr(fr, 1, 0, 3.5);
chkFr(fr, 0, 1, 2.2); // Group 2.2, mean is 4.5
chkFr(fr, 1, 1, 4.5);
chkFr(fr, 0, 7, 2.8); // Group 2.8, mean is 5.043, largest group
chkFr(fr, 1, 7, 5.042857142857143);
chkFr(fr, 0, 22, 4.4); // Group 4.4, mean is 1.5, last group
chkFr(fr, 1, 22, 1.5);
} finally {
if (fr != null) fr.delete();
Keyed.remove(Key.make("hex"));
}
}
@Test
public void testChicago() {
Frame weather = null, crimes = null, census = null;
String oldtz = Exec.exec("(getTimeZone)").getStr();
try {
weather = parse_test_file(Key.make("weather.hex"), "smalldata/chicago/chicagoAllWeather.csv");
crimes =
parse_test_file(Key.make("crimes.hex"), "smalldata/chicago/chicagoCrimes10k.csv.zip");
String fname = "smalldata/chicago/chicagoCensus.csv";
File f = find_test_file(fname);
assert f != null && f.exists() : " file not found: " + fname;
NFSFileVec nfs = NFSFileVec.make(f);
ParseSetup ps = ParseSetup.guessSetup(new Key[] {nfs._key}, false, 1);
ps.getColumnTypes()[1] = Vec.T_ENUM;
census = ParseDataset.parse(Key.make("census.hex"), new Key[] {nfs._key}, true, ps);
census =
exec_str(
"(colnames= census.hex [0 1 2 3 4 5 6 7 8] [\"Community.Area.Number\" \"COMMUNITY.AREA.NAME\" \"PERCENT.OF.HOUSING.CROWDED\" \"PERCENT.HOUSEHOLDS.BELOW.POVERTY\" \"PERCENT.AGED.16..UNEMPLOYED\" \"PERCENT.AGED.25..WITHOUT.HIGH.SCHOOL.DIPLOMA\" \"PERCENT.AGED.UNDER.18.OR.OVER.64\" \"PER.CAPITA.INCOME.\" \"HARDSHIP.INDEX\"])",
"census.hex");
crimes =
exec_str(
"(colnames= crimes.hex [0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21] [\"ID\" \"Case.Number\" \"Date\" \"Block\" \"IUCR\" \"Primary.Type\" \"Description\" \"Location.Description\" \"Arrest\" \"Domestic\" \"Beat\" \"District\" \"Ward\" \"Community.Area\" \"FBI.Code\" \"X.Coordinate\" \"Y.Coordinate\" \"Year\" \"Updated.On\" \"Latitude\" \"Longitude\" \"Location\"])",
"crimes.hex");
exec_str("(setTimeZone \"Etc/UTC\")", null);
crimes =
exec_str(
"(colnames= (= crimes.hex (tmp= unary_op_6 (day (tmp= nary_op_5 (as.Date (cols crimes.hex [2]) \"%m/%d/%Y %I:%M:%S %p\")))) [22] [0:9999]) 22 \"Day\")",
"crimes.hex");
crimes =
exec_str(
"(colnames= (= crimes.hex (tmp= binary_op_31 (+ (tmp= unary_op_7 (month nary_op_5)) #1)) [23] [0:9999]) 23 \"Month\")",
"crimes.hex");
Keyed.remove(Key.make("nary_op_30"));
crimes =
exec_str(
"(colnames= (= crimes.hex (tmp= binary_op_32 (+ (tmp= binary_op_9 (- (tmp= unary_op_8 (year nary_op_5)) #1900)) #1900)) [17] [0:9999]) 17 \"Year\")",
"crimes.hex");
crimes =
exec_str(
"(colnames= (= crimes.hex (tmp= unary_op_10 (week nary_op_5)) [24] [0:9999]) 24 \"WeekNum\")",
"crimes.hex");
Keyed.remove(Key.make("binary_op_32"));
Keyed.remove(Key.make("binary_op_31"));
Keyed.remove(Key.make("unary_op_8"));
checkSaneFrame();
crimes =
exec_str(
"(colnames= (= crimes.hex (tmp= unary_op_11 (dayOfWeek nary_op_5)) [25] [0:9999]) 25 \"WeekDay\")",
"crimes.hex");
Keyed.remove(
Key.make(
"nfs:\\C:\\Users\\cliffc\\Desktop\\h2o-3\\smalldata\\chicago\\chicagoCrimes10k.csv.zip"));
crimes =
exec_str(
"(colnames= (= crimes.hex (tmp= unary_op_12 (hour nary_op_5)) [26] [0:9999]) 26 \"HourOfDay\")",
"crimes.hex");
crimes =
exec_str(
"(colnames= (= crimes.hex (tmp= nary_op_16 (ifelse (tmp= binary_op_15 (| (tmp= binary_op_13 (== unary_op_11 \"Sun\")) (tmp= binary_op_14 (== unary_op_11 \"Sat\")))) 1 0)) [27] [0:9999]) 27 \"Weekend\")",
"crimes.hex");
// Season is incorrectly assigned in the original chicago demo; picks up the Weekend flag
crimes =
exec_str(
"(colnames= (= crimes.hex nary_op_16 [28] [0:9999]) 28 \"Season\")", "crimes.hex");
// Standard "head of 10 rows" pattern for printing
Frame subset_33 = exec_str("(rows crimes.hex [0:10])", "subset_33");
Keyed.remove(Key.make("subset_33"));
Keyed.remove(Key.make("subset_33"));
Keyed.remove(Key.make("unary_op_29"));
Keyed.remove(Key.make("nary_op_28"));
Keyed.remove(Key.make("nary_op_27"));
Keyed.remove(Key.make("nary_op_26"));
Keyed.remove(Key.make("binary_op_25"));
Keyed.remove(Key.make("binary_op_24"));
Keyed.remove(Key.make("binary_op_23"));
Keyed.remove(Key.make("binary_op_22"));
Keyed.remove(Key.make("binary_op_21"));
Keyed.remove(Key.make("binary_op_20"));
Keyed.remove(Key.make("binary_op_19"));
Keyed.remove(Key.make("binary_op_18"));
Keyed.remove(Key.make("binary_op_17"));
Keyed.remove(Key.make("nary_op_16"));
Keyed.remove(Key.make("binary_op_15"));
Keyed.remove(Key.make("binary_op_14"));
Keyed.remove(Key.make("binary_op_13"));
Keyed.remove(Key.make("unary_op_12"));
Keyed.remove(Key.make("unary_op_11"));
Keyed.remove(Key.make("unary_op_10"));
Keyed.remove(Key.make("binary_op_9"));
Keyed.remove(Key.make("unary_op_8"));
Keyed.remove(Key.make("unary_op_7"));
Keyed.remove(Key.make("unary_op_6"));
Keyed.remove(Key.make("nary_op_5"));
checkSaneFrame();
// Standard "head of 10 rows" pattern for printing
Frame subset_34 = exec_str("(rows crimes.hex [0:10])", "subset_34");
Keyed.remove(Key.make("subset_34"));
census =
exec_str(
"(colnames= census.hex [0 1 2 3 4 5 6 7 8] [\"Community.Area\" \"COMMUNITY.AREA.NAME\" \"PERCENT.OF.HOUSING.CROWDED\" \"PERCENT.HOUSEHOLDS.BELOW.POVERTY\" \"PERCENT.AGED.16..UNEMPLOYED\" \"PERCENT.AGED.25..WITHOUT.HIGH.SCHOOL.DIPLOMA\" \"PERCENT.AGED.UNDER.18.OR.OVER.64\" \"PER.CAPITA.INCOME.\" \"HARDSHIP.INDEX\"])",
"census.hex");
Keyed.remove(Key.make("subset_34"));
Frame subset_35 = exec_str("(cols crimes.hex [-3])", "subset_35");
Frame subset_36 = exec_str("(cols weather.hex [-1])", "subset_36");
subset_36 =
exec_str(
"(colnames= subset_36 [0 1 2 3 4 5] [\"Month\" \"Day\" \"Year\" \"maxTemp\" \"meanTemp\" \"minTemp\"])",
"subset_36");
crimes.remove();
weather.remove();
// nary_op_37 = merge( X Y ); Vecs in X & nary_op_37 shared
Frame nary_op_37 = exec_str("(merge subset_35 census.hex FALSE FALSE)", "nary_op_37");
// nary_op_38 = merge( nary_op_37 subset_36); Vecs in nary_op_38 and nary_pop_37 and X shared
Frame subset_41 =
exec_str(
"(rows (tmp= nary_op_38 (merge nary_op_37 subset_36 TRUE FALSE)) (tmp= binary_op_40 (<= (tmp= nary_op_39 (h2o.runif nary_op_38 30792152736.5179)) #0.8)))",
"subset_41");
// Standard "head of 10 rows" pattern for printing
Frame subset_44 = exec_str("(rows subset_41 [0:10])", "subset_44");
Keyed.remove(Key.make("subset_44"));
Keyed.remove(Key.make("subset_44"));
Keyed.remove(Key.make("binary_op_40"));
Keyed.remove(Key.make("nary_op_37"));
Frame subset_43 =
exec_str("(rows nary_op_38 (tmp= binary_op_42 (> nary_op_39 #0.8)))", "subset_43");
// Chicago demo continues on past, but this is all I've captured for now
checkSaneFrame();
} finally {
Exec.exec(
"(setTimeZone \""
+ oldtz
+ "\")"); // Restore time zone (which is global, and will affect following tests)
if (weather != null) weather.remove();
if (crimes != null) crimes.remove();
if (census != null) census.remove();
for (String s :
new String[] {
"nary_op_5",
"unary_op_6",
"unary_op_7",
"unary_op_8",
"binary_op_9",
"unary_op_10",
"unary_op_11",
"unary_op_12",
"binary_op_13",
"binary_op_14",
"binary_op_15",
"nary_op_16",
"binary_op_17",
"binary_op_18",
"binary_op_19",
"binary_op_20",
"binary_op_21",
"binary_op_22",
"binary_op_23",
"binary_op_24",
"binary_op_25",
"nary_op_26",
"nary_op_27",
"nary_op_28",
"unary_op_29",
"binary_op_30",
"binary_op_31",
"binary_op_32",
"subset_33",
"subset_34",
"subset_35",
"subset_36",
"nary_op_37",
"nary_op_38",
"nary_op_39",
"binary_op_40",
"subset_41",
"binary_op_42",
"subset_43",
"subset_44",
}) Keyed.remove(Key.make(s));
}
}