def comapare_dataframes(
cls,
df1: DataFrame,
df2: DataFrame,
excluded_keys: Union[List, str, None] = []) -> bool:
"""
Compares 2 DataFrames for exact match\
internally it use pandas.testing.assert_frame_equal
:param df1: processed data
:type df1: DataFrame
:param df2: gold standard expected data
:type df2: DataFrame
:return: True
:param excluded_keys: columns to be excluded from comparision, optional
:type excluded_keys: Union[List, str, None]
:rtype: Boolean
:raises: AssertionError Dataframe mismatch
"""
excluded_keys = excluded_keys if type(excluded_keys) == list else [
excluded_keys
]
df1 = df1.drop(*excluded_keys)
df2 = df2.drop(*excluded_keys)
sort_columns = [cols[0] for cols in df1.dtypes]
df1_sorted = df1.toPandas().sort_values(by=sort_columns,
ignore_index=True)
df2_sorted = df2.toPandas().sort_values(by=sort_columns,
ignore_index=True)
assert_frame_equal(df1_sorted, df2_sorted)
return True
def create_input_widgets(years: DataFrame, countries: DataFrame, ratings: DataFrame, dbutils: DBUtils):
min_year = years.toPandas().values[0][0]
max_year = years.toPandas().values[0][1]
country_list = list(map(lambda x: x[0], countries.toPandas().values.tolist()))
rating_list = list(map(lambda x: x[0], ratings.toPandas().values.tolist()))
# country_list.append("All")
rating_list.remove(None)
rating_list.sort()
dbutils.widgets.dropdown("year", str(min_year), list(map(str, range(min_year, max_year + 1))), "Select year")
dbutils.widgets.dropdown("country", country_list[0], country_list, "Select country")
dbutils.widgets.dropdown("rating", "C", rating_list, "Select rating")
def plot_defaults_per_country(df: DataFrame):
year = dbutils.widgets.get("year") # noqa: F821
rating = dbutils.widgets.get("rating") # noqa: F821
if len(df.head(1)) == 0:
return
ax = sns.barplot(x="Country", y="Defaults", data=df.toPandas())
ax.set_title(f"Defaults per Country of {rating} rating during {year}")
return display(ax) # noqa: F821
def plot_defaults_per_month(df: DataFrame):
year = dbutils.widgets.get("year") # noqa: F821
country = dbutils.widgets.get("country") # noqa: F821
rating = dbutils.widgets.get("rating") # noqa: F821
if len(df.head(1)) == 0:
return
ax = sns.barplot(x="Month", y="Defaults", data=df.toPandas())
ax.set_title(f"Defaults per Month in {year} in {country} of {rating} rating")
return display(ax) # noqa: F821
