def test_smvConcatHist(self):
df = self.createDF("k:String;v:String", "a,1;b,2")
res = dfhelper(df)._smvConcatHist("k", "v")
self.assertEqual(res, """Histogram of k_v: String sort by Key
key count Pct cumCount cumPct
a_1 1 50.00% 1 50.00%
b_2 1 50.00% 2 100.00%
-------------------------------------------------""")
def test_smvHistInt(self):
df = self.createDF("k:String;v:Integer", "a,1;b,2")
res = dfhelper(df)._smvHist("v")
self.assertEqual(res, """Histogram of v: Numeric sort by Key
key count Pct cumCount cumPct
1.0 1 50.00% 1 50.00%
2.0 1 50.00% 2 100.00%
-------------------------------------------------""")
def test_smvHistDate(self):
df = self.createDF("k:Date;v:Integer", "2010-01-01,1;2010-01-02,2")
res = dfhelper(df)._smvHist("k")
self.assertEqual(res, """Histogram of k: String sort by Key
key count Pct cumCount cumPct
2010-01-01 1 50.00% 1 50.00%
2010-01-02 1 50.00% 2 100.00%
-------------------------------------------------""")
def test_smvConcatHist(self):
df = self.createDF("k:String;v:String", "a,1;b,2")
res = dfhelper(df)._smvConcatHist("k", "v")
self.assertEqual(res, """Histogram of k_v: String sort by Key
key count Pct cumCount cumPct
a_1 1 50.00% 1 50.00%
b_2 1 50.00% 2 100.00%
-------------------------------------------------""")
def test_smvHist(self):
df = self.createDF("k:String;v:Integer", "a,1;b,2")
res = dfhelper(df)._smvHist("k")
self.assertEqual(res, """Histogram of k: String sort by Key
key count Pct cumCount cumPct
a 1 50.00% 1 50.00%
b 1 50.00% 2 100.00%
-------------------------------------------------""")
def test_smvBinHist(self):
import smv.helpers as smv
df = self.createDF("k:String;v:Integer", "a,10;b,200;a,30")
res = dfhelper(df)._smvBinHist(("v", 100))
self.assertEqual(res, """Histogram of v: with BIN size 100.0
key count Pct cumCount cumPct
0.0 2 66.67% 2 66.67%
200.0 1 33.33% 3 100.00%
-------------------------------------------------""")
def test_smvCountHist(self):
import smv.helpers as smv
df = self.createDF("k:String;v:String", "a,1;b,2;a,3")
res = dfhelper(df)._smvCountHist(["k"], 1)
self.assertEqual(res, """Histogram of N_k: with BIN size 1.0
key count Pct cumCount cumPct
1.0 1 50.00% 1 50.00%
2.0 1 50.00% 2 100.00%
-------------------------------------------------""")
def test_smvFreqHistDate(self):
import smv.helpers as smv
df = self.createDF("k:Date;v:String", "2010-01-01,1;2010-01-02,2;2010-01-02,3")
res = dfhelper(df)._smvFreqHist("k")
self.assertEqual(res, """Histogram of k: String sorted by Frequency
key count Pct cumCount cumPct
2010-01-02 2 66.67% 2 66.67%
2010-01-01 1 33.33% 3 100.00%
-------------------------------------------------""")
def test_smvFreqHist(self):
import smv.helpers as smv
df = self.createDF("k:String;v:String", "a,1;b,2;a,3")
res = dfhelper(df)._smvFreqHist("k")
self.assertEqual(res, """Histogram of k: String sorted by Frequency
key count Pct cumCount cumPct
a 2 66.67% 2 66.67%
b 1 33.33% 3 100.00%
-------------------------------------------------""")
def test_smvBinHist(self):
import smv.helpers as smv
df = self.createDF("k:String;v:Integer", "a,10;b,200;a,30")
res = dfhelper(df)._smvBinHist(("v", 100))
self.assertEqual(res, """Histogram of v: with BIN size 100.0
key count Pct cumCount cumPct
0.0 2 66.67% 2 66.67%
200.0 1 33.33% 3 100.00%
-------------------------------------------------""")
def test_smvCountHist(self):
import smv.helpers as smv
df = self.createDF("k:String;v:String", "a,1;b,2;a,3")
res = dfhelper(df)._smvCountHist(["k"], 1)
self.assertEqual(res, """Histogram of N_k: with BIN size 1.0
key count Pct cumCount cumPct
1.0 1 50.00% 1 50.00%
2.0 1 50.00% 2 100.00%
-------------------------------------------------""")
def test_smvFreqHist(self):
import smv.helpers as smv
df = self.createDF("k:String;v:String", "a,1;b,2;a,3")
res = dfhelper(df)._smvFreqHist("k")
self.assertEqual(res, """Histogram of k: String sorted by Frequency
key count Pct cumCount cumPct
a 2 66.67% 2 66.67%
b 1 33.33% 3 100.00%
-------------------------------------------------""")
def test_smvHistDate(self):
df = self.createDF("k:Date;v:Integer", "2010-01-01,1;2010-01-02,2")
res = dfhelper(df)._smvHist("k")
self.assertEqual(
res, """Histogram of k: String sort by Key
key count Pct cumCount cumPct
2010-01-01 1 50.00% 1 50.00%
2010-01-02 1 50.00% 2 100.00%
-------------------------------------------------""")
def test_smvFreqHistDate(self):
import smv.helpers as smv
df = self.createDF("k:Date;v:String",
"2010-01-01,1;2010-01-02,2;2010-01-02,3")
res = dfhelper(df)._smvFreqHist("k")
self.assertEqual(
res, """Histogram of k: String sorted by Frequency
key count Pct cumCount cumPct
2010-01-02 2 66.67% 2 66.67%
2010-01-01 1 33.33% 3 100.00%
-------------------------------------------------""")
def test_smvEdd(self):
df = self.createDF("k:String;v:Integer", "a,1;b,2")
res = dfhelper(df)._smvEdd()
self.assertEqual(res, """k Non-Null Count 2
k Null Count 0
k Min Length 1
k Max Length 1
k Approx Distinct Count 2
v Non-Null Count 2
v Null Count 0
v Average 1.5
v Standard Deviation 0.7071067811865476
v Min 1.0
v Max 2.0""")
def test_smvEdd(self):
df = self.createDF("k:String;v:Integer", "a,1;b,2")
res = dfhelper(df)._smvEdd()
self.assertEqual(res, """k Non-Null Count 2
k Null Count 0
k Min Length 1
k Max Length 1
k Approx Distinct Count 2
v Non-Null Count 2
v Null Count 0
v Average 1.5
v Standard Deviation 0.7071067811865476
v Min 1.0
v Max 2.0""")
