def test_indices():
config = sett.SetContainer(test_run_config, model_config)
data = pd.read_csv(test_csv).set_index('block_year')
for break_window in ['1Year', '2Year', '3Year']:
print break_window
for past_yr in range(1, 7):
print past_yr
pX_train, Y_train, pX_valid, Y_valid, pX_test, Y_test, dic_year = \
train_valid_test_split(
data,
break_window,
config.static_features,
config.cv_cuts[
'thirty_seventy'],
config.past,
config.future,
past_yr=past_yr,
config=config)
assert np.all(pX_train.index == Y_train.index)
assert np.all(pX_valid.index == Y_valid.index)
assert np.all(pX_test.index == Y_test.index)
assert len(pX_valid.ix[pX_train.index, :].dropna()) == 0
assert len(pX_valid.ix[pX_train.index, :].dropna()) == 0
assert len(pX_test.ix[pX_train.index, :].dropna()) == 0
assert len(pX_valid.ix[pX_train.index, :].dropna()) == 0
def test_balance_set():
"""
Given a set of labels of a training set
balance_set will determine how balanced the
set is and then downsample to a specfied proportion
Input
-----
fraction_1s: float
fraction of ones we want
fraction_0s: float
fraction of zeros we want
Output
------
labels: ls
index of labels that have the proper portion to
downsample
"""
config = sett.SetContainer(test_run_config, model_config)
#data = get_feature_table(config.tablename)
data = pd.read_csv(test_csv).set_index('block_year')
break_window = '2Year'
pX_train, Y_train, pX_valid, Y_valid, pX_test, Y_test, date_dic = train_valid_test_split(
data,
break_window,
config.static_features,
config.cv_cuts['thirty_seventy'],
config.past,
config.future,
past_yr=4)
ls_balance = [[0.3, 0.7], [0.2, 0.8], [0.1, 0.9], [0.5, 0.5]]
for balance in ls_balance:
break_bal, nobreak_bal = balance
X_bal, Y_bal = downsample(pX_train,
Y_train,
downsample_balance=balance,
Verbose=True)
print 'Y_train: ', np.sum(Y_train == 1)
print check_balance(Y_train)
balance_after = check_balance(Y_bal)
assert np.isclose(break_bal, balance_after['break'],
atol=1e-4), '{} {}'.format(break_bal,
balance_after['break'])
assert np.isclose(nobreak_bal, balance_after['no_break'], atol=1e-4)
print balance_after
def test_1window():
config = sett.SetContainer(test_run_config, model_config)
data = pd.read_csv(test_csv).set_index('block_year')
label = '1Year'
train_yrs, valid_yrs, test_yrs = train_valid_test_split(data,
label,
config.static_features,
config.cv_cuts[
'thirty_seventy'],
config.past,
config.future,
testing=True)
train_yrs = np.array(train_yrs, dtype=int)
valid_yrs = np.array(valid_yrs, dtype=int)
test_yrs = np.array(test_yrs, dtype=int)
assert np.all(np.array(range(2004, 2011)) == train_yrs)
assert np.all(np.array(range(2011, 2014)) == valid_yrs)
assert 2014 == test_yrs
def test_1roving_window():
config = sett.SetContainer(test_run_config, model_config)
data = pd.read_csv(test_csv).set_index('block_year')
label = '1Year'
start_yr = 2004
for _past_yr in range(1, 7):
train_yrs, valid_yrs, test_yrs = train_valid_test_split(data,
label,
config.static_features,
config.cv_cuts[
'thirty_seventy'],
config.past,
config.future,
past_yr=_past_yr,
testing=True)
train_yrs = np.array(train_yrs, dtype=int)
valid_yrs = np.array(valid_yrs, dtype=int)
test_yrs = np.array(test_yrs, dtype=int)
assert train_yrs[0] == (start_yr - 1) + _past_yr
assert valid_yrs[-1] == 2013
assert 2014 == test_yrs
def test_3window():
config = sett.SetContainer(test_run_config, model_config)
data = pd.read_csv(test_csv).set_index('block_year')
label = '3Year'
train_yrs, valid_yrs, test_yrs = train_valid_test_split(data,
label,
config.static_features,
config.cv_cuts[
'thirty_seventy'],
config.past,
config.future,
past_yr=0,
testing=True,
Verbose=True)
train_yrs = np.array(train_yrs, dtype=int)
valid_yrs = np.array(valid_yrs, dtype=int)
print(train_yrs)
assert np.all(np.array(range(2006, 2010)) == train_yrs), train_yrs
assert np.all(np.array(range(2010, 2012)) == valid_yrs), valid_yrs
assert 2012 == test_yrs