feat: add new alert description (#1724)

fabclmnt · azory-ydata · web-flow · commit fc8dfe92fa52 · 2025-03-10T20:25:43.000-07:00
* fix: add new alert description

* fix(linting): code formatting

* feat: add new dirty categories config

* feat: informative banner for end-users

* fix(linting): code formatting

* fix: rename the alert for Dirty category

* fix(linting): code formatting

* fix: linter fixes

* fix(linting): code formatting

* chore: set dirty categories config to False

* chore: disable linter for expected code design

* fix(linting): code formatting

---------

Co-authored-by: Azory YData Bot &lt;azory@ydata.ai&gt;
diff --git a/src/ydata_profiling/config.py b/src/ydata_profiling/config.py
@@ -68,6 +68,8 @@ class CatVars(BaseModel):
     redact: bool = False
     histogram_largest: int = 50
     stop_words: List[str] = []
+    dirty_categories: bool = False
+    dirty_categories_threshold: float = 0.85
 
 
 class BoolVars(BaseModel):
diff --git a/src/ydata_profiling/model/alerts.py b/src/ydata_profiling/model/alerts.py
@@ -52,6 +52,9 @@ class AlertType(Enum):
     DUPLICATES = auto()
     """This variable contains duplicates."""
 
+    NEAR_DUPLICATES = auto()
+    """This variable contains duplicates."""
+
     SKEWED = auto()
     """This variable is highly skewed."""
 
@@ -70,6 +73,9 @@ class AlertType(Enum):
     UNIQUE = auto()
     """This variable has unique values."""
 
+    DIRTY_CATEGORY = auto()
+    """This variable is a categories with potential fuzzy values, and for that reason might incur in consistency issues."""
+
     CONSTANT_LENGTH = auto()
     """This variable has a constant length."""
 
@@ -205,7 +211,29 @@ def _get_description(self) -> str:
         if self.values is not None:
             return f"Dataset has {self.values['n_duplicates']} ({fmt_percent(self.values['p_duplicates'])}) duplicate rows"
         else:
-            return "Dataset has duplicated values"
+            return "Dataset has no duplicated rows"
+
+
+class NearDuplicatesAlert(Alert):
+    def __init__(
+        self,
+        values: Optional[Dict] = None,
+        column_name: Optional[str] = None,
+        is_empty: bool = False,
+    ):
+        super().__init__(
+            alert_type=AlertType.NEAR_DUPLICATES,
+            values=values,
+            column_name=column_name,
+            fields={"n_near_dups"},
+            is_empty=is_empty,
+        )
+
+    def _get_description(self) -> str:
+        if self.values is not None:
+            return f"Dataset has {self.values['n_near_dups']} ({fmt_percent(self.values['p_near_dups'])}) near duplicate rows"
+        else:
+            return "Dataset has no near duplicated rows"
 
 
 class EmptyAlert(Alert):
@@ -249,6 +277,28 @@ def _get_description(self) -> str:
             return f"[{self.column_name}] has a high cardinality"
 
 
+class DirtyCategoryAlert(Alert):
+    def __init__(
+        self,
+        values: Optional[Dict] = None,
+        column_name: Optional[str] = None,
+        is_empty: bool = False,
+    ):
+        super().__init__(
+            alert_type=AlertType.DIRTY_CATEGORY,
+            values=values,
+            column_name=column_name,
+            fields={"n_fuzzy_vals"},
+            is_empty=is_empty,
+        )
+
+    def _get_description(self) -> str:
+        if self.values is not None:
+            return f"[{self.column_name}] has {self.values['n_fuzzy_vals']} fuzzy values: {fmt_percent(self.values['p_fuzzy_vals'])} per category"
+        else:
+            return f"[{self.column_name}] no dirty categories values."
+
+
 class HighCorrelationAlert(Alert):
     def __init__(
         self,
diff --git a/src/ydata_profiling/model/pandas/describe_categorical_pandas.py b/src/ydata_profiling/model/pandas/describe_categorical_pandas.py
@@ -16,6 +16,7 @@
     series_handle_nulls,
     series_hashable,
 )
+from ydata_profiling.utils.information import DisplayInfo
 
 
 def get_character_counts_vc(vc: pd.Series) -> pd.Series:
@@ -210,6 +211,9 @@ def length_summary_vc(vc: pd.Series) -> dict:
     return summary
 
 
+_displayed_catvar_banner = False
+
+
 @describe_categorical_1d.register
 @series_hashable
 @series_handle_nulls
@@ -226,6 +230,8 @@ def pandas_describe_categorical_1d(
     Returns:
         A dict containing calculated series description values.
     """
+    # Global info banner
+    global _displayed_catvar_banner
 
     # Make sure we deal with strings (Issue #100)
     series = series.astype(str)
@@ -262,4 +268,13 @@ def pandas_describe_categorical_1d(
     if config.vars.cat.words:
         summary.update(word_summary_vc(value_counts, config.vars.cat.stop_words))
 
+    if config.vars.cat.dirty_categories:  # noqa: SIM102
+        if not _displayed_catvar_banner:
+            display_info = DisplayInfo(
+                title="Identify dirty categories with ydata-sdk",
+                info_text="This feature is only available for ydata-sdk users. Register to give try it.",
+            )
+            display_info.display_message()
+            _displayed_catvar_banner = True
+
     return config, series, summary
diff --git a/src/ydata_profiling/report/presentation/flavours/html/templates/alerts/alert_dirty_category.html b/src/ydata_profiling/report/presentation/flavours/html/templates/alerts/alert_dirty_category.html
@@ -0,0 +1 @@
+<a href="#pp_var_{{ alert.anchor_id }}"><code>{{ alert.column_name }}</code></a> has dirty categories: {{ alert.values['n_fuzzy_vals'] }} distinct values
diff --git a/src/ydata_profiling/report/presentation/flavours/html/templates/alerts/alert_near_duplicates.html b/src/ydata_profiling/report/presentation/flavours/html/templates/alerts/alert_near_duplicates.html
@@ -0,0 +1 @@
+Dataset has {{ alert.values['n_near_dups'] }} ({{ alert.values['p_near_dups'] | fmt_percent }}) <a href="#near_duplicate">near duplicate rows</a>
diff --git a/src/ydata_profiling/utils/information.py b/src/ydata_profiling/utils/information.py
@@ -21,22 +21,48 @@ def in_jupyter_notebook() -> bool:
     return isiPython
 
 
+class DisplayInfo:
+    def __init__(
+        self,
+        title: str,
+        info_text: str,
+        link: str = "ttps://ydata.ai/register",
+    ):
+        self.title = title
+        self.link = link
+        self.info_text = info_text
+
+    def display_message(self) -> None:
+        """
+        Display an HTML message in case the user is in a Jupyter Notebook
+        """
+        if in_jupyter_notebook():
+            from IPython.display import HTML, display
+
+            info = f"""
+            <div>
+                <ins><a href="{self.link}">{self.title}</a></ins>
+                <p>
+                    {self.info_text}
+                </p>
+            </div>
+            """
+            display(HTML(info))
+        else:
+            info = (
+                f"\033[1;34m{self.title}\033[0m"
+                + "\n"
+                + f"{self.info_text}"
+                + "\n"
+                + f"Register at {self.link}"
+            )
+            print(info)  # noqa: T201
+
+
 def display_banner() -> None:
     global _displayed_banner
-    if in_jupyter_notebook() and not _displayed_banner:
-        from IPython.display import HTML, display
-
-        banner_html = f"""
-        <div>
-            <ins><a href="{link}">{title}</a></ins>
-            <p>
-                {info_text}
-            </p>
-        </div>
-        """
-        display(HTML(banner_html))
-    else:
-        print(f"\033[1;34m{title}\033[0m")  # noqa: T201
-        print(info_text)  # noqa: T201
-        print(f"Register at {link}")  # noqa: T201
+    banner_info = DisplayInfo(title=title, info_text=info_text)
+
+    if not _displayed_banner:
+        banner_info.display_message()
         _displayed_banner = True

Original file line number	Diff line number	Diff line change
`@@ -0,0 +1 @@`
	`1`	`+<a href="#pp_var_{{ alert.anchor_id }}"><code>{{ alert.column_name }}</code></a> has dirty categories: {{ alert.values['n_fuzzy_vals'] }} distinct values`
Original file line number	Diff line number	Diff line change
`@@ -0,0 +1 @@`
	`1`	`+Dataset has {{ alert.values['n_near_dups'] }} ({{ alert.values['p_near_dups'] \| fmt_percent }}) <a href="#near_duplicate">near duplicate rows</a>`