def test_direction_feature():
df = pd.DataFrame([{
'open': 1,
'high': 9,
'low': 1,
'close': 1,
'volume': 100
}, {
'open': 1,
'high': 8,
'low': 2,
'close': 2,
'volume': 100
}, {
'open': 1,
'high': 7,
'low': 3,
'close': 3,
'volume': 100
}, {
'open': 1,
'high': 6,
'low': 2,
'close': 4,
'volume': 100
}, {
'open': 1,
'high': 5,
'low': 1,
'close': 5,
'volume': 100
}, {
'open': 1,
'high': 4,
'low': 0,
'close': 6,
'volume': 100
}, {
'open': 1,
'high': 3,
'low': 0,
'close': 7,
'volume': 100
}])
direction_close_producer = DirectionFeatureProducer()
close_direction_result = direction_close_producer.produce(
df)['Y_close_direction_1']
expected_close_direction = [1] * len(df)
expected_close_direction[-1] = np.nan
array_equal(close_direction_result, expected_close_direction)
direction_low_producer = DirectionFeatureProducer(feature='low', lags=2)
low_2_direction_result = direction_low_producer.produce(
df)['Y_low_direction_2']
expected_low_direction = [1, 1, 0, 0, 0, np.nan, np.nan]
array_equal(low_2_direction_result, expected_low_direction)
def test_fi_feature():
ohlc_df = pd.DataFrame([{
'open': 1,
'high': 9,
'low': 1,
'close': 1,
'volume': 100
}, {
'open': 1,
'high': 8,
'low': 1,
'close': 2,
'volume': 100
}, {
'open': 1,
'high': 7,
'low': 1,
'close': 4,
'volume': 100
}, {
'open': 1,
'high': 6,
'low': 1,
'close': 8,
'volume': 100
}, {
'open': 1,
'high': 5,
'low': 1,
'close': 16,
'volume': 100
}, {
'open': 1,
'high': 4,
'low': 1,
'close': 32,
'volume': 100
}, {
'open': 1,
'high': 3,
'low': 1,
'close': 64,
'volume': 100
}])
fi_producer = ForceIndexFeatureProducer(period=3)
fi_result = fi_producer.produce(ohlc_df)
assert 'fi_1' in fi_result
assert 'fi_3' in fi_result
expect_fi_1 = [np.nan, 200, 400, 800, 1600, 3200]
array_equal(fi_result['fi_1'].values, expect_fi_1)
def test_typical_price():
ohlc_df = pd.DataFrame([{
'open': 1,
'high': 1,
'low': 1,
'close': 1,
'volume': 100
}, {
'open': 1,
'high': 3,
'low': 1,
'close': 2,
'volume': 100
}, {
'open': 1,
'high': 7,
'low': 1,
'close': 4,
'volume': 100
}])
typical_price_producer = TypicalPriceFeatureProducer()
typical_price_result = typical_price_producer.produce(ohlc_df)['tp'].values
expected_typical_price = [1, 2, 4]
assert array_equal(typical_price_result, expected_typical_price)
def test_pct_change_producer():
ohlc_df = pd.DataFrame([
{'open': 1, 'high': 9, 'low': 1, 'close': 1, 'volume': 100},
{'open': 1, 'high': 8, 'low': 1, 'close': 2, 'volume': 100},
{'open': 1, 'high': 7, 'low': 1, 'close': 4, 'volume': 100},
{'open': 1, 'high': 6, 'low': 1, 'close': 8, 'volume': 100},
{'open': 1, 'high': 5, 'low': 1, 'close': 16, 'volume': 100},
{'open': 1, 'high': 4, 'low': 1, 'close': 32, 'volume': 100},
{'open': 1, 'high': 3, 'low': 1, 'close': 64, 'volume': 100}
])
close_pct_change_producer = PercentChangeFeatureProducer()
result_close_pct_change = close_pct_change_producer.produce(ohlc_df)['close_pct_change'].values
expected_close_pct_change = [1] * len(result_close_pct_change)
expected_close_pct_change[0] = np.nan
assert array_equal(result_close_pct_change, expected_close_pct_change)
volume_pct_change_producer = PercentChangeFeatureProducer(feature='volume')
result_vol_pct_change = volume_pct_change_producer.produce(ohlc_df)['volume_pct_change'].values
expected_vol_pct_change = [0] * len(result_vol_pct_change)
expected_vol_pct_change[0] = np.nan
assert array_equal(result_vol_pct_change, expected_vol_pct_change)
def test_bb_feature():
ohlc_df = pd.DataFrame([
{'open': 1, 'high': 9, 'low': 1, 'close': 1, 'volume': 100},
{'open': 1, 'high': 8, 'low': 1, 'close': 2, 'volume': 100},
{'open': 1, 'high': 7, 'low': 1, 'close': 4, 'volume': 100},
{'open': 1, 'high': 6, 'low': 1, 'close': 8, 'volume': 100},
{'open': 1, 'high': 5, 'low': 1, 'close': 16, 'volume': 100},
{'open': 1, 'high': 4, 'low': 1, 'close': 32, 'volume': 100},
{'open': 1, 'high': 3, 'low': 1, 'close': 64, 'volume': 100}
])
bb_producer = BollingerBandsFeatureProducer(period=3)
bb_result = bb_producer.produce(ohlc_df)
sma_producer = SimpleMovingAverageFeatureProducer(period=3)
sma_result = sma_producer.produce(ohlc_df)
assert 'bb_3_middle' in bb_result
assert 'bb_3_upper' in bb_result
assert 'bb_3_lower' in bb_result
array_equal(bb_result['bb_3_middle'].values, sma_result['close_sma_3'].values)
def test_simple_moving_average():
ohlc_df = pd.DataFrame([{
'open': 1,
'high': 9,
'low': 1,
'close': 1,
'volume': 100
}, {
'open': 1,
'high': 8,
'low': 1,
'close': 2,
'volume': 100
}, {
'open': 1,
'high': 7,
'low': 1,
'close': 3,
'volume': 100
}, {
'open': 1,
'high': 6,
'low': 1,
'close': 4,
'volume': 100
}, {
'open': 1,
'high': 5,
'low': 1,
'close': 5,
'volume': 100
}, {
'open': 1,
'high': 4,
'low': 1,
'close': 6,
'volume': 100
}, {
'open': 1,
'high': 3,
'low': 1,
'close': 7,
'volume': 100
}])
close_2_sma_provider = SimpleMovingAverageFeatureProducer(period=2)
close_2_sma_result = close_2_sma_provider.produce(
ohlc_df)['close_sma_2'].values
expect_close_2_sma = [np.nan, 1.5, 2.5, 3.5, 4.5, 5.5, 6.5]
assert array_equal(close_2_sma_result, expect_close_2_sma)
def test_exponential_moving_average():
ohlc_df = pd.DataFrame([{
'open': 1,
'high': 9,
'low': 1,
'close': 1,
'volume': 100
}, {
'open': 1,
'high': 8,
'low': 1,
'close': 2,
'volume': 100
}, {
'open': 1,
'high': 7,
'low': 1,
'close': 3,
'volume': 100
}, {
'open': 1,
'high': 6,
'low': 1,
'close': 4,
'volume': 100
}, {
'open': 1,
'high': 5,
'low': 1,
'close': 5,
'volume': 100
}, {
'open': 1,
'high': 4,
'low': 1,
'close': 6,
'volume': 100
}, {
'open': 1,
'high': 3,
'low': 1,
'close': 7,
'volume': 100
}])
volume_2_ema_provider = ExponentialMovingAverageFeatureProducer(
period=2, feature='volume')
volume_2_ema_result = volume_2_ema_provider.produce(
ohlc_df)['volume_ema_2'].values
expect_volume_2_ema = [100] * len(volume_2_ema_result)
assert array_equal(volume_2_ema_result, expect_volume_2_ema)
def test_time_lags_producer():
ohlc_df = pd.DataFrame([
{'open': 1, 'high': 9, 'low': 1, 'close': 1, 'volume': 100},
{'open': 1, 'high': 8, 'low': 1, 'close': 2, 'volume': 100},
{'open': 1, 'high': 7, 'low': 1, 'close': 3, 'volume': 100},
{'open': 1, 'high': 6, 'low': 1, 'close': 4, 'volume': 100},
{'open': 1, 'high': 5, 'low': 1, 'close': 5, 'volume': 100},
{'open': 1, 'high': 4, 'low': 1, 'close': 6, 'volume': 100},
{'open': 1, 'high': 3, 'low': 1, 'close': 7, 'volume': 100}
])
close_3_producer = TimeLagsFeatureProducer(lags=3, feature='close')
close_3_result = close_3_producer.produce(ohlc_df)['close_tlag_3'].values
expect_result_close_3 = np.asarray([np.nan, np.nan, np.nan, 1., 2., 3., 4.])
assert array_equal(close_3_result, expect_result_close_3)