def work2(df, f):
setname = "sp500"
taname = "base1"
(phase1, phase2, phase3) = \
feat_select.split_dates(feat_select.load_feat(taname, setname))
df2 = feat_select.apply(df, phase2, "label5", "_p2")
df2 = feat_select.apply(df2, phase3, "label5", "_p3")
feat_select.ana2(df2, f, setname)
def work2(df, f):
setname = "sp500"
taname = "base1"
(phase1, phase2, phase3) = \
feat_select.split_dates(feat_select.load_feat(taname, setname))
df2 = feat_select.apply(df,phase2, "label5", "_p2")
df2 = feat_select.apply(df2,phase3, "label5", "_p3")
feat_select.ana2(df2, f, setname)
def phase1_dump(taname, setname):
dfTa = feat_select.load_feat(taname, setname)
(phase1, phase2, phase3) = feat_select.split_dates(dfTa)
dfmetas = feat_select.flat_metas(feat_select.get_metas(phase1))
outdir = os.path.join(root, "data", "feat_select", "phase1_dump")
if not os.path.exists(outdir):
os.makedirs(outdir)
dfmetas.to_pickle(os.path.join(outdir, "%s_%s.pkl" % (setname, taname)))
def phase1_dump(taname, setname):
dfTa = feat_select.load_feat(taname, setname)
(phase1, phase2, phase3) = feat_select.split_dates(dfTa)
dfmetas = feat_select.flat_metas(feat_select.get_metas(phase1))
outdir = os.path.join(root, "data", "feat_select", "phase1_dump")
if not os.path.exists(outdir):
os.makedirs(outdir)
dfmetas.to_pickle(os.path.join(outdir, "%s_%s.pkl" % (setname, taname)))
def work(df, f):
for i in range(10):
frm = 50 * i
to = frm + 50
setname = "sp500R%dT%d" % (frm, to)
taname = "base1"
(phase1, phase2, phase3) = \
feat_select.split_dates(feat_select.load_feat(taname, setname))
df2 = feat_select.apply(df, phase2, "label5", "_p2")
df2 = feat_select.apply(df2, phase3, "label5", "_p3")
feat_select.ana2(df2, f, setname)
def work(df, f):
for i in range(10):
frm = 50 * i
to = frm + 50
setname = "sp500R%dT%d" % (frm, to)
taname = "base1"
(phase1, phase2, phase3) = \
feat_select.split_dates(feat_select.load_feat(taname, setname))
df2 = feat_select.apply(df,phase2, "label5", "_p2")
df2 = feat_select.apply(df2,phase3, "label5", "_p3")
feat_select.ana2(df2, f, setname)
orig_direct_n_set = abs_direct_n_set.copy()
print len(abs_direct_p_set)
print len(abs_direct_n_set)
print >> f, "=" * 8
for i in range(10):
frm = 50 * i
to = frm + 50
setname = "sp500R%dT%d" % (frm, to)
taname = "base1"
filename = os.path.join(
dataroot, "phase1_dump", "sp500_base1_apply_phase1_%s_%s_%d.pkl" %
(setname, taname, args.depth))
if not os.path.exists(filename):
df2 = feat_select.apply(
df,
feat_select.split_dates(feat_select.load_feat(taname,
setname))[0],
"label5", "_p1")
df2.to_pickle(filename)
df2 = pd.read_pickle(filename)
feat_select.ana_apply(df2, "_p1", setname, f)
cur_p_set = set(df2[df2.direct_p1 == 1].name.unique())
cur_n_set = set(df2[df2.direct_p1 == -1].name.unique())
abs_direct_p_set = abs_direct_p_set.intersection(cur_p_set)
abs_direct_n_set = abs_direct_n_set.intersection(cur_n_set)
print list(abs_direct_n_set)
df.loc[:,"istable"] = df.apply(lambda row: 1 if row["name"] in abs_direct_p_set else \
(1 if row["name"] in abs_direct_n_set else 0), axis = 1)
df.loc[:, "direct"] = df.apply(lambda row: 0
if row["istable"] == 0 else row["direct"],
axis=1)
