public void TestSignaturesV2_3_X()
{
Column col = _df["name"];
col = _df["age"];
DataFrame df = _df.ToDF();
df = df.ToDF("name2", "age2");
StructType schema = _df.Schema();
Assert.NotNull(schema);
_df.PrintSchema();
_df.Explain();
_df.Explain(true);
_df.Explain(false);
Assert.Equal(2, _df.Columns().ToArray().Length);
_df.IsLocal();
_df.IsStreaming();
using (var tempDir = new TemporaryDirectory())
{
// The following is required for *CheckPoint().
_spark.SparkContext.SetCheckpointDir(tempDir.Path);
_df.Checkpoint();
_df.Checkpoint(false);
_df.LocalCheckpoint();
_df.LocalCheckpoint(false);
}
_df.WithWatermark("time", "10 minutes");
_df.Show();
_df.Show(10);
_df.Show(10, 10);
_df.Show(10, 10, true);
_df.Join(_df);
_df.Join(_df, "name");
_df.Join(_df, new[] { "name" });
_df.Join(_df, new[] { "name" }, "outer");
_df.Join(_df, _df["age"] == _df["age"]);
_df.Join(_df, _df["age"] == _df["age"], "outer");
_df.CrossJoin(_df);
_df.SortWithinPartitions("age");
_df.SortWithinPartitions("age", "name");
_df.SortWithinPartitions();
_df.SortWithinPartitions(_df["age"]);
_df.SortWithinPartitions(_df["age"], _df["name"]);
_df.Sort("age");
_df.Sort("age", "name");
_df.Sort();
_df.Sort(_df["age"]);
_df.Sort(_df["age"], _df["name"]);
_df.OrderBy("age");
_df.OrderBy("age", "name");
_df.OrderBy();
_df.OrderBy(_df["age"]);
_df.OrderBy(_df["age"], _df["name"]);
_df.Hint("broadcast");
_df.Hint("broadcast", new[] { "hello", "world" });
_df.Col("age");
_df.ColRegex("age");
_df.As("alias");
_df.Alias("alias");
_df.Select("age");
_df.Select("age", "name");
_df.Select();
_df.Select(_df["age"]);
_df.Select(_df["age"], _df["name"]);
_df.SelectExpr();
_df.SelectExpr("age * 2");
_df.SelectExpr("age * 2", "abs(age)");
_df.Filter(_df["age"] > 21);
_df.Filter("age > 21");
_df.Where(_df["age"] > 21);
_df.Where("age > 21");
_df.GroupBy("age");
_df.GroupBy("age", "name");
_df.GroupBy();
_df.GroupBy(_df["age"]);
_df.GroupBy(_df["age"], _df["name"]);
_df.Rollup("age");
_df.Rollup("age", "name");
_df.Rollup();
_df.Rollup(_df["age"]);
_df.Rollup(_df["age"], _df["name"]);
_df.Cube("age");
_df.Cube("age", "name");
_df.Cube();
_df.Cube(_df["age"]);
_df.Cube(_df["age"], _df["name"]);
_df.Agg(Avg(_df["age"]));
_df.Agg(Avg(_df["age"]), Avg(_df["name"]));
_df.Limit(10);
_df.Union(_df);
_df.UnionByName(_df);
_df.Intersect(_df);
_df.Except(_df);
_df.Sample(0.5);
_df.Sample(0.5, true);
_df.Sample(0.5, false, 12345);
_df.RandomSplit(new[] { 0.2, 0.8 });
_df.RandomSplit(new[] { 0.2, 0.8 }, 12345);
_df.WithColumn("age2", _df["age"]);
_df.WithColumnRenamed("age", "age2");
_df.Drop();
_df.Drop("age");
_df.Drop("age", "name");
_df.Drop(_df["age"]);
_df.DropDuplicates();
_df.DropDuplicates("age");
_df.DropDuplicates("age", "name");
_df.Describe();
_df.Describe("age");
_df.Describe("age", "name");
_df.Summary();
_df.Summary("count");
_df.Summary("count", "mean");
_df.Head(2);
_df.Head();
_df.First();
_df.Take(3).ToArray();
_df.Collect().ToArray();
_df.ToLocalIterator().ToArray();
_df.Count();
_df.Repartition(2);
_df.Repartition(2, _df["age"]);
_df.Repartition(_df["age"]);
_df.Repartition();
_df.RepartitionByRange(2, _df["age"]);
_df.RepartitionByRange(_df["age"]);
_df.Coalesce(1);
_df.Distinct();
_df.Persist();
_df.Cache();
_df.Unpersist();
_df.CreateTempView("view");
_df.CreateOrReplaceTempView("view");
_df.CreateGlobalTempView("global_view");
_df.CreateOrReplaceGlobalTempView("global_view");
}
public void Describe_Test01()
{
var mlContext = new MLContext();
var dict = new Dictionary <string, List <object> >
{
{ "product_id", new List <object>()
{
1, 1, 2, 2, 2, 2, 2
} },
{ "retail_price", new List <object>()
{
2, 2, 5, 5, 5, 5, 5
} },
{ "quantity", new List <object>()
{
1, 2, 4, 8, 16, 32, 64
} },
{ "city", new List <object>()
{
"SF", "SJ", "SF", "SJ", "Miami", "Orlando", "SJ"
} },
{ "state", new List <object>()
{
"CA", "CA", "CA", "CA", "FL", "FL", "PR"
} },
};
var df = new DataFrame(dict);
var descDf = df.Describe(false, "product_id", "quantity", "state");
Assert.True(descDf.RowCount() == 11);
Assert.True(descDf.ColCount() == 3);
var str = descDf.ToStringBuilder();
Assert.Equal("Count", descDf.Index[0]);
Assert.Equal(7, descDf["product_id", 0]);
Assert.Equal(7, descDf["quantity", 0]);
Assert.Equal(7, descDf["state", 0]);
Assert.Equal("Unique", descDf.Index[1]);
Assert.Equal(2, descDf["product_id", 1]);
Assert.Equal(7, descDf["quantity", 1]);
Assert.Equal(3, descDf["state", 1]);
Assert.Equal("Top", descDf.Index[2]);
Assert.Equal(2, descDf["product_id", 2]);
Assert.Equal(1, descDf["quantity", 2]);
Assert.Equal("CA", descDf["state", 2]);
Assert.Equal("Freq", descDf.Index[3]);
Assert.Equal(5, descDf["product_id", 3]);
Assert.Equal(1, descDf["quantity", 3]);
Assert.Equal(4, descDf["state", 3]);
Assert.Equal("Mean", descDf.Index[4]);
Assert.Equal(1.714286, descDf["product_id", 4]);
Assert.Equal(18.142857, descDf["quantity", 4]);
Assert.Equal(DataFrame.NAN, descDf["state", 4]);
Assert.Equal("Std", descDf.Index[5]);
Assert.Equal(0.487950, descDf["product_id", 5]);
Assert.Equal(22.937804, descDf["quantity", 5]);
Assert.Equal(DataFrame.NAN, descDf["state", 5]);
Assert.Equal("Min", descDf.Index[6]);
Assert.Equal(1, descDf["product_id", 6]);
Assert.Equal(1, descDf["quantity", 6]);
Assert.Equal(DataFrame.NAN, descDf["state", 6]);
Assert.Equal("25%", descDf.Index[7]);
Assert.Equal(1.5, descDf["product_id", 7]);
Assert.Equal(3.0, descDf["quantity", 7]);
Assert.Equal(DataFrame.NAN, descDf["state", 7]);
Assert.Equal("Median", descDf.Index[8]);
Assert.Equal(2.0, descDf["product_id", 8]);
Assert.Equal(8.0, descDf["quantity", 8]);
Assert.Equal(DataFrame.NAN, descDf["state", 8]);
Assert.Equal("75%", descDf.Index[9]);
Assert.Equal(2d, descDf["product_id", 9]);
Assert.Equal(24d, descDf["quantity", 9]);
Assert.Equal(DataFrame.NAN, descDf["state", 9]);
Assert.Equal("Max", descDf.Index[10]);
Assert.Equal(2, descDf["product_id", 10]);
Assert.Equal(64, descDf["quantity", 10]);
Assert.Equal(DataFrame.NAN, descDf["state", 10]);
}
