def process_pandas(data, endog_idx=0, exog_idx=None, index_idx=None):
names = data.columns
if isinstance(endog_idx, int):
endog_name = names[endog_idx]
endog = data[endog_name].copy()
if exog_idx is None:
exog = data.drop([endog_name], axis=1)
else:
exog = data[names[exog_idx]].copy()
else:
endog = data.loc[:, endog_idx].copy()
endog_name = list(endog.columns)
if exog_idx is None:
exog = data.drop(endog_name, axis=1)
elif isinstance(exog_idx, int):
exog = data[names[exog_idx]].copy()
else:
exog = data[names[exog_idx]].copy()
if index_idx is not None: # NOTE: will have to be improved for dates
index = Index(data.iloc[:, index_idx])
endog.index = index
exog.index = index.copy()
data = data.set_index(names[index_idx])
exog_name = list(exog.columns)
dataset = Dataset(data=data, names=list(names), endog=endog,
exog=exog, endog_name=endog_name, exog_name=exog_name)
return dataset
def process_pandas(data, endog_idx=0, exog_idx=None, index_idx=None):
names = data.columns
if isinstance(endog_idx, (int, long)):
endog_name = names[endog_idx]
endog = data[endog_name].copy()
if exog_idx is None:
exog = data.drop([endog_name], axis=1)
else:
exog = data[names[exog_idx]].copy()
else:
endog = data.loc[:, endog_idx].copy()
endog_name = list(endog.columns)
if exog_idx is None:
exog = data.drop(endog_name, axis=1)
elif isinstance(exog_idx, (int, long)):
exog = data[names[exog_idx]].copy()
else:
exog = data[names[exog_idx]].copy()
if index_idx is not None: # NOTE: will have to be improved for dates
index = Index(data.iloc[:, index_idx])
endog.index = index
exog.index = index.copy()
data = data.set_index(names[index_idx])
exog_name = list(exog.columns)
dataset = Dataset(data=data, names=list(names), endog=endog,
exog=exog, endog_name=endog_name, exog_name=exog_name)
return dataset
def sort_columns_based_first_character(columns: pd.Index,
list_order: List[str]) -> np.ndarray:
"""
columns の 先頭文字をlist_order の順番に沿って並び替える
Params::
columns: 対象のcolumn
list_order: 先頭文字の順番
"""
columns = columns.copy()
df_col = pd.DataFrame(columns.values, columns=["colname"])
df_col["order"] = np.nan # None で初期化する. ここに順番を表す数字を入力する
# 先頭文字が長い順にsortして、長い方から当てはめていく
df = pd.DataFrame(list_order, columns=["first_character"])
df["length"] = df["first_character"].apply(lambda x: len(x))
df = df.sort_values(by=["length"], ascending=False)
for i in df.index:
first_character = df.loc[i, "first_character"]
boolwk = (df_col["colname"].str.contains("^" + first_character)
& df_col["order"].isna())
df_col.loc[boolwk, "order"] = i
val = df_col["order"].fillna(
-1).max() # dtype=object で 1, nan, None の列のmax()は空になる
df_col["order"] = df_col["order"].fillna(val + 1)
return df_col.sort_values(by=["order", "colname"])["colname"].values
def nanmap(columns: pd.Index, dict_map: dict) -> np.ndarray:
"""
columns.map(dict) では dict に定義されていない変数は nan になる.
定義されてないcolumnはそのまま残すようにする
"""
columns = columns.copy()
columnswk = columns.map(dict_map).fillna("__work").values
columns = columns.values
columns[(columnswk != "__work")] = columnswk[(columnswk != "__work")]
return columns
def from_pandas_index(cls, index: pd.Index, dim: Hashable):
from .variable import IndexVariable
if index.name is None:
name = dim
index = index.copy()
index.name = dim
else:
name = index.name
data = PandasIndexingAdapter(index)
index_var = IndexVariable(dim, data, fastpath=True)
return cls(index, dim), {name: index_var}
