def plot_2d_line(df: pd.DataFrame,
x: str,
y: str,
color: str = None,
criteria: str = None) -> go.Figure:
"""
:param df: Data set
:param x: name of column X
:param y: name of column Y
:param color: name of grouping column
:param criteria: python condition
:return: plot figure
"""
if criteria is not None:
df = filter_records(df, criteria)
if not x in df.columns:
logger.log_error('x not found in columns')
return
if not y in df.columns:
logger.log_error('y not found in columns')
return
if not color in df.columns and not color is None:
logger.log_warn('color column not found')
color = None
return px.line(df, x=x, y=y, color=color)
def drop_na(df: pd.DataFrame, columns='all') -> pd.DataFrame:
df_copy = df.copy()
if columns is 'all':
columns = df.columns
n = len(columns)
columns = [x for x in columns if x in df.columns]
if n is not len(columns):
logger.log_warn('some columns not found in dataFrame')
df_copy = df_copy.dropna(subset=columns)
return df_copy
def fill_na_values(df: pd.DataFrame, columns='all', value=0) -> pd.DataFrame:
df_copy = df.copy()
if columns is 'all':
df_copy = df_copy.fillna(value)
else:
n = len(columns)
columns = [x for x in columns if x in df.columns]
if n is not len(columns):
logger.log_warn('some columns not found in dataFrame')
df_copy[columns] = df_copy[columns].fillna(value)
return df_copy
def drop_na(df: pd.DataFrame, columns='all') -> pd.DataFrame:
"""
:param df: Data Set
:param columns: default : "all"
:return: new Data Set
"""
df_copy = df.copy()
if columns is 'all':
columns = df.columns
n = len(columns)
columns = [x for x in columns if x in df.columns]
if n is not len(columns):
logger.log_warn('some columns not found in dataFrame')
df_copy = df_copy.dropna(subset=columns)
return df_copy
def plot_2d(df: pd.DataFrame, x: str, y: str, color: str = None, trendline: bool = False, criteria: str = None) -> go.Figure:
if criteria is not None:
df = filter_records(df, criteria)
if not x in df.columns:
logger.log_error('x not found in columns')
return
if not y in df.columns:
logger.log_error('y not found in columns')
return
if not color in df.columns and not color is None:
logger.log_warn('color column not found')
color = None
if trendline:
trendline = 'ols'
return px.scatter(df, x=x, y=y, color=color, trendline=trendline)
def fill_na_values(df: pd.DataFrame, columns='all', value=0) -> pd.DataFrame:
"""
:param df: Data Set
:param columns: default: "all"
:param value: replace value default: 0
:return: new Data Set
"""
df_copy = df.copy()
if columns is 'all':
df_copy = df_copy.fillna(value)
else:
n = len(columns)
columns = [x for x in columns if x in df.columns]
if n is not len(columns):
logger.log_warn('some columns not found in dataFrame')
df_copy[columns] = df_copy[columns].fillna(value)
return df_copy