def setUp(self):
self.rs1 = []
for year in range(2001, 2011):
r = Row()
r.yyyy = year
self.rs1.append(r)
self.rs1 = PRows(self.rs1, 'yyyy')
self.rs2 = []
start_month = 200101
for i in range(36):
r = Row()
r.yyyymm = yyyymm(start_month, i)
self.rs2.append(r)
self.rs2 = PRows(self.rs2, 'yyyymm')
self.rs3 = []
start_date = 20010101
for i in range(30):
r = Row()
r.yyyymmdd = yyyymmdd(start_date, i)
self.rs3.append(r)
self.rs3 = PRows(self.rs3, 'yyyymmdd')
with dbopen(':memory:') as c:
c.save('indport.csv')
# to pseudo monthly data
rs = []
for rs1 in c.reel('indport order by date', group=lambda r: str(r.date)[0:4]):
for r in rs1:
r.yyyy = int(str(r.date)[0:4])
r.fcode = 'A' + str(r.date)[4:]
del r.date
rs.append(r)
self.indport = PRows(rs, 'yyyy', 'fcode')
rs = []
for rs1 in c.reel('indport order by date', group=lambda r: str(r.date)[0:6]):
for r in rs1:
r.yyyymm = int(str(r.date)[0:6])
r.fcode = 'A' + str(r.date)[6:]
del r.date
rs.append(r)
self.indport1 = PRows(rs, 'yyyymm', 'fcode')
class TestPRows(unittest.TestCase):
def setUp(self):
self.rs1 = []
for year in range(2001, 2011):
r = Row()
r.yyyy = year
self.rs1.append(r)
self.rs1 = PRows(self.rs1, 'yyyy')
self.rs2 = []
start_month = 200101
for i in range(36):
r = Row()
r.yyyymm = yyyymm(start_month, i)
self.rs2.append(r)
self.rs2 = PRows(self.rs2, 'yyyymm')
self.rs3 = []
start_date = 20010101
for i in range(30):
r = Row()
r.yyyymmdd = yyyymmdd(start_date, i)
self.rs3.append(r)
self.rs3 = PRows(self.rs3, 'yyyymmdd')
with dbopen(':memory:') as c:
c.save('indport.csv')
# to pseudo monthly data
rs = []
for rs1 in c.reel('indport order by date', group=lambda r: str(r.date)[0:4]):
for r in rs1:
r.yyyy = int(str(r.date)[0:4])
r.fcode = 'A' + str(r.date)[4:]
del r.date
rs.append(r)
self.indport = PRows(rs, 'yyyy', 'fcode')
rs = []
for rs1 in c.reel('indport order by date', group=lambda r: str(r.date)[0:6]):
for r in rs1:
r.yyyymm = int(str(r.date)[0:6])
r.fcode = 'A' + str(r.date)[6:]
del r.date
rs.append(r)
self.indport1 = PRows(rs, 'yyyymm', 'fcode')
def test_indi_sort(self):
with self.assertRaises(ValueError):
# there are not enough element to make portfolios in 2009
self.indport.pn('cnsmr', 2).pn('manuf', 3).pavg('other')
avgport = self.indport.where(lambda r: r.yyyy < 2009)\
.pn('cnsmr', 2).pn('manuf', 3).pavg('other')
self.assertEqual(avgport[0].n, 76)
self.assertEqual(avgport[1].n, 45)
self.assertEqual(avgport[2].n, 3)
self.assertEqual(avgport[3].n, 7)
self.assertEqual(avgport[4].n, 37)
self.assertEqual(avgport[5].n, 80)
self.assertEqual(round(avgport[0].other, 2), -0.63)
indport = self.indport.where(lambda r: r.yyyy < 2009).pn('cnsmr', 10)
other1 = []
for year in range(2001, 2009):
other1.append(mean0(indport.where(lambda r: r.pn_cnsmr == 1 and r.yyyy == year)['other']))
self.assertEqual(other1, [-1.249, -1.418, -0.838, -0.98, -0.944, -1.027, -1.75, -4.143])
other10 = []
for year in range(2001, 2009):
other10.append(mean0(indport.where(lambda r: r.pn_cnsmr == 10 and r.yyyy == year)['other']))
self.assertEqual(other10, [1.415, 1.486, 1.235, 0.96, 1.062, 1.174, 1.34, 4.014])
pat = self.indport.where(lambda r: r.yyyy < 2009)\
.pn('cnsmr', 10).pavg('other', pncols='pn_cnsmr').pat('pn_cnsmr')
self.assertEqual(round(st.mean(other10) - st.mean(other1), 2), float(pat.lines[0][11][:4]))
indport = self.indport.where(lambda r: r.yyyy < 2009).pn('cnsmr', 2).pn('manuf', 3)
# pavg.show()
other21 = []
other23 = []
for year in range(2001, 2009):
pavg1 = indport.where(lambda r: r.pn_cnsmr == 2 and r.pn_manuf == 1 and r.yyyy == year)['other']
pavg2 = indport.where(lambda r: r.pn_cnsmr == 2 and r.pn_manuf == 3 and r.yyyy == year)['other']
other21.append(st.mean(pavg1))
other23.append(st.mean(pavg2))
pat = indport.pavg('other').pat().lines
self.assertEqual(round(st.mean(other21), 3), float(pat[2][1].split()[0]))
self.assertEqual(round(st.mean(other23), 3), float(pat[2][3].split()[0]))
self.assertEqual(round(st.mean(other23) - st.mean(other21), 3), float(pat[2][4][:5]))
indport.pavg('other', pncols='pn_cnsmr, pn_manuf').pat().csv()
def test_indi_sort2(self):
"weighted average"
avgport = self.indport.where(lambda r: r.yyyy <= 2015).pn('cnsmr', 10)
hlth = avgport.where(lambda r: r.yyyy == 2001 and r.pn_cnsmr == 3)['hlth']
other = avgport.where(lambda r: r.yyyy == 2001 and r.pn_cnsmr == 3)['other']
total = sum(hlth)
result = []
for x, y in zip(other, hlth):
result.append(x * y / total)
self.assertEqual(sum(result),
avgport.pavg('other', 'hlth')\
.where(lambda r: r.yyyy == 2001 and r.pn_cnsmr == 3)['other'][0])
def test_indi_sort3(self):
def fn(rs):
n = round(len(rs) / 2)
return [rs[:n], rs[n:]]
self.assertEqual(self.indport.pn('cnsmr', 2).pavg('other').pat().lines,
self.indport.pn('cnsmr', fn).pavg('other').pat().lines)
def test_dpn(self):
avgport = self.indport.pn('cnsmr', 4).dpn('manuf', 3, 'hlth', 2).pavg('other')
for r in avgport.where(lambda r: r.yyyy < 2016):
self.assertTrue(r.n == 10 or r.n == 11)
seq1 = avgport.where(lambda r: r.pn_cnsmr == 3 and r.pn_manuf == 1 and r.pn_hlth == 2)['other']
seq2 = avgport.where(lambda r: r.pn_cnsmr == 3 and r.pn_manuf == 3 and r.pn_hlth == 2)['other']
pat = avgport.pat().lines
self.assertEqual(round(st.mean(seq1), 3), float(pat[14][2].split()[0]))
self.assertEqual(round(st.mean(seq2), 3), float(pat[16][2].split()[0]))
self.assertEqual(round(st.mean(seq2) - st.mean(seq1), 3), float(pat[17][2][:5]))
def test_pnroll(self):
a = self.indport.between(2003).dpnroll(5, 'cnsmr', 5, 'manuf', 4)
for rs in a.roll(5, 5):
for rs1 in rs.order('fcode, yyyy').group('fcode'):
self.assertTrue(same(rs1['pn_cnsmr, pn_manuf']))
xs = a.pavg('other')
self.assertEqual(round(xs[0].other, 3), -1.013)
self.assertEqual(round(xs[-1].other, 3), -0.426)
a = self.indport.between(2003).pnroll(5, 'cnsmr', 5, 'manuf', 4, 'hi_tec', 3)
for rs in a.roll(5, 5):
for rs1 in rs.order('fcode, yyyy').group('fcode'):
self.assertTrue(same(rs1['pn_cnsmr, pn_manuf, pn_hi_tec']))
with self.assertRaises(Exception):
a.pavg('other')
a = self.indport.between(2003).pnroll(5, 'cnsmr', 5, 'manuf', 4)
# independent sort raises exception because there is not enough elements
with self.assertRaises(Exception):
a.pavg('other')
a = self.indport.between(2003).pnroll(5, 'cnsmr', 2, 'manuf', 2)
xs = a.pavg('other')
self.assertEqual(round(xs[0].other, 3), -0.338)
self.assertEqual(round(xs[-1].other, 3), -0.014)
def test_pns(self):
self.indport.dpn('cnsmr', 4, 'manuf', 3, 'hlth', 2)
self.assertEqual(len(self.indport.pncols), 3)
self.indport.pncols.clear()
self.indport.dpn('cnsmr', 2, 'manuf', 3)
self.assertEqual(len(self.indport.pncols), 2)
def test_famac(self):
fit = self.indport.famac('other ~ cnsmr + manuf + hi_tec + hlth')
self.assertEqual(round(fit[0].intercept, 2), 0.02)
self.assertEqual(round(fit[0].cnsmr, 2), 0.44)
self.assertEqual(round(fit[0].manuf, 2), 0.16)
self.assertEqual(round(fit[0].hi_tec, 2), 0.03)
self.assertEqual(round(fit[0].hlth, 2), 0.10)
fitavg = fit.tsavg()
for var, val in zip(['cnsmr', 'manuf', 'hi_tec', 'hlth'], fitavg.lines[1][2:]):
self.assertEqual(mean1(fit[var]), val)
def test_rollover(self):
lengths = []
for rs0 in self.rs1.roll(3, 2):
lengths.append(len(rs0))
self.assertEqual(lengths, [3, 3, 3, 3, 2])
lengths = []
for rs0 in self.rs2.where(lambda r: r.yyyymm > 200103).roll(12, 12):
lengths.append(len(rs0))
self.assertEqual(lengths, [12, 12, 9])
lengths = []
for rs0 in self.rs2.where(lambda r: r.yyyymm > 200103).roll(24, 12):
lengths.append(len(rs0))
self.assertEqual(lengths, [24, 21, 9])
lengths = []
for rs0 in self.rs3.roll('2 weeks', '1 week'):
lengths.append(len(rs0))
self.assertEqual(lengths, [14, 14, 14, 9, 2])