def assemble_prediction_frame(frames: Dict[str, pd.DataFrame]):
# filter non frames
valid_frames = {
head: frame.copy()
for head, frame in frames.items() if frame is not None
}
for head, frame in valid_frames.items():
frame.columns = pd.MultiIndex.from_product([[head],
frame.columns.to_list()])
# join all frames and keep the order of the passed dictionary
df = pd.concat(valid_frames.values(), axis=1, join='inner', copy=False)
# monkey patch prediction frame
df.map_prediction_to_target = lambda: map_prediction_to_target(
df, PREDICTION_COLUMN_NAME, TARGET_COLUMN_NAME)
return df
def assemble_prediction_frame(frames: Dict[str, pd.DataFrame]):
# filter non frames
valid_frames = {head: (frame.as_joined_frame() if isinstance(frame, MultiFrameDecorator) else frame.copy())
for head, frame in frames.items() if frame is not None }
# convert into MultiIndex column headers
for head, frame in valid_frames.items():
if isinstance(frame, pd.Series):
frame = frame.to_frame()
valid_frames[head] = frame
frame.columns = pd.MultiIndex.from_product([[head], frame.columns.to_list()])
# join all frames and keep the order of the passed dictionary
df = pd.concat(valid_frames.values(), axis=1, join='inner', copy=False)
# monkey patch prediction frame
df.map_prediction_to_target = lambda: map_prediction_to_target(df, PREDICTION_COLUMN_NAME, TARGET_COLUMN_NAME)
return df
import numpy as np
from pandas_ml_common import pd
print('NUMPY VERSION', np.__version__)
def _with_multi_index(df, header):
df = df.copy()
df.columns = pd.MultiIndex.from_product([[header], df.columns])
return df
def _with_multi_index_row(df, header):
df = df.copy()
df.index = pd.MultiIndex.from_product([[header], df.index])
return df
DATA_PATH = os.path.join(os.path.dirname(os.path.abspath(__file__)), ".data")
TEST_FILE = os.path.join(DATA_PATH, "SPY.csv")
TEST_DF = pd.read_csv(TEST_FILE, index_col='Date', parse_dates=True)
TEST_MULTI_INDEX_DF = _with_multi_index(TEST_DF, "A").join(
_with_multi_index(TEST_DF, "B"))
TEST_MUTLI_INDEX_ROW_DF = pd.concat([
_with_multi_index_row(TEST_DF.tail(), "A"),
_with_multi_index_row(TEST_DF.tail() + 1.0, "B")
],
axis=0)