def test_na_at_i_end(test_data):
data = test_data(SHAPE)
last_i = np.shape(data)[0] - 1
data = test_data(SHAPE, last_i, 3)
actual = impy.locf(data, axis=1)
data[last_i, 3] = data[last_i - 1, 3]
assert np.array_equal(actual, data)
def pre_lof(data, year, year1):
# 设定时间序列
# data['TIMESTAMP'] = pd.to_datetime(data['TIMESTAMP'])
data = data.set_index('TIMESTAMP')
if year.strip():
if year1.strip():
data = data[year:year1]
else:
data = data[year]
# 补充了数据 ,这时补充的数据都是 0
df_period = data.resample('D').sum()
# 替换空值为其他值
df_period_clone = df_period
df_period_clone.replace(0, np.nan, inplace=True)
# 平滑后的数据
isnullcon = df_period_clone.isnull().any()
# print(isnullcon['FP_TOTALENG'])
# 这里来个判断，如果数据里没有需要平滑的点，那么直接输出
if isnullcon['FP_TOTALENG']:
df_after = impy.locf(df_period_clone, axis='FP_TOTALENG')
for i in range(len(df_period_clone)):
# print('-----平缓后数据-----')
# print(df_after[0][i])
# print('-----平缓前数据-----')
# print(df_period['FP_TOTALENG'][i])
df_period_clone['FP_TOTALENG'][i] = df_after[0][i]
return df_period_clone
def test_na_at_i_end(self):
"""Check if at row[last_i],column[i]=row[last_i-1],column[i]"""
last_i = np.shape(self.data_c)[0] - 1
self.data_c[last_i][3] = np.nan
actual = impy.locf(self.data_c, axis=1)
self.data_c[last_i][3] = self.data_c[last_i - 1][3]
self.assertTrue(np.array_equal(actual, self.data_c))
def replace_data(df_period, field):
pd.set_option('display.max_rows', None)
# 替换空值为其他值
df_period_clone = df_period
df_period_clone.replace(0, np.nan, inplace=True)
# 平滑后的数据
isnullcon = df_period_clone.isnull().any()
# print(isnullcon['FP_TOTALENG'])
# 这里来个判断，如果数据里没有需要平滑的点，那么直接输出
df_peroid_v = df_period_clone[field]
df_peroid_v.reset_index(drop=True, inplace=True)
print(df_peroid_v.index)
if isnullcon[field]:
df_after = impy.locf(df_period_clone, axis=field)
for i in range(len(df_period_clone)):
print('-----平缓后数据-----')
print(df_after[0][i])
print('-----平缓前数据-----')
print(df_period[field][i])
df_period_clone[field][i] = df_after[0][i]
# print(df_period_clone)
return df_period_clone
def test_locf_(test_data):
data = test_data(SHAPE)
imputed = impy.locf(data)
return_na_check(imputed)
def test_out_of_bounds(test_data):
"""Check out of bounds error, should throw BadInputError for any axis outside [0,1]"""
data = test_data(SHAPE)
with np.testing.assert_raises(error.BadInputError):
impy.locf(data, axis=3)
def test_na_at_i(test_data):
data = test_data(SHAPE, 3, 3)
actual = impy.locf(data, axis=1)
data[3, 3] = data[2, 3]
assert np.array_equal(actual, data)
def test_na_at_i_start(test_data):
data = test_data(SHAPE)
actual = impy.locf(data, axis=1)
data[0, 0] = data[1, 0]
assert np.array_equal(actual, data)
def test_na_at_i(self):
"""Check if a null row[i],column[j]=row[i-1],column[j]"""
self.data_c[3][3] = np.nan
actual = impy.locf(self.data_c, axis=1)
self.data_c[3][3] = self.data_c[2][3]
self.assertTrue(np.array_equal(actual, self.data_c))
def test_na_at_i_start(self):
"""Check if null value at a row[0],column[0]=row[1],column[0]"""
self.data_c[0][0] = np.nan
actual = impy.locf(self.data_c, axis=1)
self.data_c[0][0] = self.data_c[1][0]
self.assertTrue(np.array_equal(actual, self.data_c))
def test_return_type(self):
""" Check return type, should return an np.ndarray"""
imputed = impy.locf(self.data_m)
self.assertTrue(isinstance(imputed, np.ndarray))
def test_impute_missing_values(self):
""" After imputation, no NaN's should exist"""
imputed = impy.locf(self.data_m)
self.assertFalse(np.isnan(imputed).any())
