Docs clarify statistics outputs

docs/dqx/docs/guide/data_profiling.mdx

@@ -85,7 +85,7 @@ Tables can be loaded and profiled using `profile_table`.
         table="catalog1.schema1.table1",
         columns=["col1", "col2", "col3"],  # specify columns to profile
     )
-    
+
     print("Summary Statistics:", summary_stats)
     print("Generated Profiles:", profiles)
     ```
@@ -139,6 +139,33 @@ The profiler can discover and profile multiple tables in Unity Catalog. Tables c
   </TabItem>
 </Tabs>
 
+### Understanding Summary Statistics
+<Admonition type="info" title="How the data is profiled">
+To ensure efficiency, summary statistics are computed on a **sampled subset** of your data, not the entire dataset. By default, the profiler proceeds as follows:
+
+1. It **samples** the rows of the dataset you provide by using Spark’s `DataFrame.sample` (default `sample_fraction = 0.3`), which is probabilistic, hence the sampled row count is not guaranteed to be exact; the number may be slightly higher or lower.
+2. A **limit** is applied to the sampled rows (default `limit = 1000`).
+3. Statistics are computed on this final **sampled-and-limited** DataFrame.
+
+**These are defaults that can be customized programmatically as detailed in the [Profiling Options](#profiling-options) section, or via `profiler_config` in the [configuration file](/docs/installation/#configuration-file) if DQX is installed as a tool.**
+
+Rule of thumb: `count ≈ min(limit, sample_fraction × total_rows)`.
+</Admonition>
+
+#### Summary Statistics Reference
+
+| Field             | Meaning                                                                 | Notes                                                               |
+|-------------------|-------------------------------------------------------------------------|---------------------------------------------------------------------|
+| `count`           | Rows actually profiled (after sampling **and** limit)                   | `≈ min(limit, sample_fraction × total_rows)`                        |
+| `mean`            | Arithmetic average of **non-null numeric** values                       | N/A for non-numeric                                                 |
+| `stddev`          | Sample standard deviation of **non-null numeric** values                | N/A for non-numeric                                                 |
+| `min`             | Smallest **non-null** value                                             | String = lexicographic; Date/Timestamp = earliest; Numeric = minimum|
+| `25` / `50` / `75`| Approximate 25th/50th/75th percentiles of **non-null numeric** values   | Uses Spark approximate quantiles                                    |
+| `max`             | Largest **non-null** value                                              | String = lexicographic; Date/Timestamp = latest; Numeric = maximum  |
+| `count_non_null`  | Number of non-null entries within the profiled rows                     |                                                                     |
+| `count_null`      | Number of null entries within the profiled rows                         | `count_non_null + count_null = count`                               |
+
+
 ## Profiling Options
 
 The profiler supports extensive configuration options to customize the profiling behavior.
@@ -159,30 +186,30 @@ You can use `options` parameter to pass a dictionary with custom options when pr
         "sample_fraction": 0.2,       # Sample 20% of the data
         "sample_seed": 42,            # Seed for reproducible sampling
         "limit": 2000,                # Limit to 2000 records after sampling
-        
+
         # Outlier detection options
         "remove_outliers": True,      # Enable outlier detection for min/max rules
         "outlier_columns": ["price", "age"],  # Only detect outliers in specific columns
         "num_sigmas": 2.5,            # Use 2.5 standard deviations for outlier detection
-        
+
         # Null value handling
         "max_null_ratio": 0.05,       # Generate is_not_null rule if <5% nulls
-        
+
         # String handling
         "trim_strings": True,         # Trim whitespace from strings before analysis
         "max_empty_ratio": 0.02,      # Generate is_not_null_or_empty rule if <2% empty strings
-        
+
         # Distinct value analysis
         "distinct_ratio": 0.01,       # Generate is_in rule if <1% distinct values
         "max_in_count": 20,           # Maximum items in is_in rule list
-        
+
         # Value rounding
         "round": True,                # Round min/max values for cleaner rules
     }
 
     ws = WorkspaceClient()
     profiler = DQProfiler(ws)
-    
+
     # Apply custom options for profiling a DataFrame
     summary_stats, profiles = profiler.profile(input_df, options=custom_options)
 

docs/dqx/docs/installation.mdx

@@ -97,7 +97,7 @@ by setting the 'DQX_FORCE_INSTALL' environment variable. The following options a
 * `DQX_FORCE_INSTALL=global databricks labs install dqx`: will force the installation to be for root only (`/Applications/dqx`)
 * `DQX_FORCE_INSTALL=user databricks labs install dqx`: will force the installation to be for user only (`/Users/<user>/.dqx`)
 
-**Configration file**
+#### Configuration file
 
 DQX configuration file can contain multiple run configurations for different pipelines or projects defining specific input, output and quarantine locations, etc.
 The "default" run configuration is created during the installation. When DQX is upgraded, the configuration is preserved.