def filter_records(df: pd.DataFrame, criteria: str) -> pd.DataFrame:
try:
result = df[df.eval(criteria)]
return result
except Exception as e:
logger.log_error(e)
return df
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 box_plot(df: pd.DataFrame, x: str, y: str, criteria: str = None) -> go.Figure:
if criteria is not None:
df = filter_records(df, criteria)
if not pd.Series([x, y]).isin(df.columns).all():
logger.log_error('columns not founded')
return
return px.box(df, x=x, y=y)
def pie_chart(df: pd.DataFrame, r: str, theta: str, criteria: str = None) -> go.Figure:
if criteria is not None:
df = filter_records(df, criteria)
if not pd.Series([theta, r]).isin(df.columns).all():
logger.log_error('columns not founded')
return
return px.bar_polar(df, r=r, theta=theta)
def histogram(df: pd.DataFrame, x: str, bins: int = 30, criteria: str = None) -> go.Figure:
if criteria is not None:
df = filter_records(df, criteria)
if x not in df.columns:
logger.log_error('column not found')
return
return px.histogram(df, x=x, nbins=bins)
def plot_3d(df: pd.DataFrame, x: str, y: str, z: str, color: str = None, criteria: str = None) -> go.Figure:
if criteria is not None:
df = filter_records(df, criteria)
if color is None and not pd.Series([x, y, z]).isin(df.columns).all() or \
color is not None and not pd.Series([x, y, z, color]).isin(df.columns).all():
logger.log_error('columns not found')
return
return px.scatter_3d(data_frame=df, x=x, y=y, z=z, color=color)
def heatmap(df: pd.DataFrame, x: str, y: str, 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
return px.density_heatmap(df, x=x, y=y)
def filter_records(df: pd.DataFrame, criteria: str) -> pd.DataFrame:
"""
:param df: Data Set
:param criteria: python condition
:return: new Data Set
"""
try:
result = df[df.eval(criteria)]
return result
except Exception as e:
logger.log_error(e)
return df
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 box_plot(df: pd.DataFrame,
x: str,
y: str,
criteria: str = None) -> go.Figure:
"""
:param df: Data Set
:param x: name of column X
:param y: name of column Y
:param criteria: python condition
:return: plot figure
"""
if criteria is not None:
df = filter_records(df, criteria)
if not pd.Series([x, y]).isin(df.columns).all():
logger.log_error('columns not founded')
return
return px.box(df, x=x, y=y)
def histogram(df: pd.DataFrame,
x: str,
bins: int = 30,
criteria: str = None) -> go.Figure:
"""
:param df: Data Set
:param x: column name for plotting (must be numeric)
:param bins: data bins
:param criteria: python condition
:return: histogram figure
"""
if criteria is not None:
df = filter_records(df, criteria)
if x not in df.columns:
logger.log_error('column not found')
return
return px.histogram(df, x=x, nbins=bins)
def plot_3d(df: pd.DataFrame,
x: str,
y: str,
z: 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 z: name of column Z
:param color: name of grouping column
:param criteria: python condition
:return: plot figure
"""
if criteria is not None:
df = filter_records(df, criteria)
if color is None and not pd.Series([x, y, z]).isin(df.columns).all() or \
color is not None and not pd.Series([x, y, z, color]).isin(df.columns).all():
logger.log_error('columns not found')
return
return px.scatter_3d(data_frame=df, x=x, y=y, z=z, color=color)