Add sorting of the summary statistics table

CHANGES.rst

@@ -34,8 +34,8 @@ Minor changes
 
 * A "stats" panel has been added to the TableReport, showing summary statistics
   for all columns (number of missing values, mean, etc. -- similar to
-  ``pandas.info()`` ) in a table.
-  :pr:`1056` by :user:`Jérôme Dockès <jeromedockes>`.
+  ``pandas.info()`` ) in a table. It can be sorted by each column.
+  :pr:`1056` and :pr:`1068` by :user:`Jérôme Dockès <jeromedockes>`.
 
 * The selection in the TableReport's sample table can now be manipulated with
   the keyboard. :pr:`1065` by :user:`Jérôme Dockès <jeromedockes>`.

examples/00_getting_started.py

@@ -2,7 +2,6 @@
 Getting Started
 ===============
 
-
 This guide showcases the features of ``skrub``, an open-source package that aims at
 bridging the gap between tabular data sources and machine-learning models.
 

skrub/_reporting/_data/templates/report.css

@@ -5,6 +5,7 @@
 {% include "copybutton.css" %}
 {% include "column-summaries.css" %}
 {% include "dataframe-sample.css" %}
+{% include "summary-statistics.css" %}
 {% include "tabs.css" %}
 {% include "tooltip.css" %}
 {% include "column-filter.css" %}

skrub/_reporting/_data/templates/report.js

@@ -558,6 +558,72 @@ if (customElements.get('skrub-table-report') === undefined) {
     }
     SkrubTableReport.register(TabList);
 
+    class sortableTable extends Manager {
+        constructor(elem, exchange) {
+            super(elem, exchange);
+            this.elem.querySelectorAll("button[data-role='sort-button']").forEach(
+                b => b.addEventListener("click", e => this.sort(e)));
+        }
+
+        getVal(row, tableColIdx) {
+            const td = row.querySelectorAll("td")[tableColIdx];
+            if (!td.hasAttribute("data-value")) {
+                return td.textContent;
+            }
+            let value = td.dataset.value;
+            if (td.hasAttribute("data-numeric")) {
+                value = Number(value);
+            }
+            return value;
+        }
+
+        compare(rowA, rowB, tableColIdx, ascending) {
+            let valA = this.getVal(rowA, tableColIdx);
+            let valB = this.getVal(rowB, tableColIdx);
+            // NaNs go at the bottom regardless of sorting order
+            if(typeof(valA) === "number" && typeof(valB) === "number"){
+                if(isNaN(valA) && !isNaN(valB)){
+                    return 1;
+                }
+                if(isNaN(valB) && !isNaN(valA)){
+                    return -1;
+                }
+            }
+            // When the values are equal, keep the original dataframe column
+            // order
+            if (!(valA > valB || valB > valA)) {
+                valA = Number(rowA.dataset.dataframeColumnIdx);
+                valB = Number(rowB.dataset.dataframeColumnIdx);
+                return valA - valB;
+            }
+            // Sort
+            if (!ascending) {
+                [valA, valB] = [valB, valA];
+            }
+            return valA > valB ? 1 : -1;
+        }
+
+        sort(event) {
+            const colHeaders = Array.from(this.elem.querySelectorAll("thead tr th"));
+            const tableColIdx = colHeaders.indexOf(event.target.closest("th"));
+            const body = this.elem.querySelector("tbody");
+            const rows = Array.from(body.querySelectorAll("tr"));
+            const ascending = event.target.dataset.direction === "ascending";
+
+            rows.sort((a, b) => this.compare(a, b, tableColIdx, ascending));
+
+            this.elem.querySelectorAll("button").forEach(b => b.removeAttribute("data-is-active"));
+            event.target.dataset.isActive = "";
+
+            body.innerHTML = "";
+            for (let r of rows) {
+                body.appendChild(r);
+            }
+        }
+
+    }
+    SkrubTableReport.register(sortableTable);
+
     class SelectedColumnsDisplay extends Manager {
 
         constructor(elem, exchange) {

skrub/_reporting/_data/templates/summary-statistics.css

@@ -0,0 +1,58 @@
+.summary-stats-table {
+    margin: 2px;
+}
+
+
+th.sort-button-group-wrapper {
+    --btn-width: 2rem;
+    --btn-group-width: calc(var(--btn-width) * 2);
+    position: relative;
+    padding-top: var(--micro);
+    padding-bottom: var(--micro);
+    padding-right: calc(var(--tiny) + var(--btn-group-width));
+}
+
+.sort-button-group {
+    position: absolute;
+    top: -1px;
+    bottom: 0;
+    right: calc(-1 * var(--btn-group-width));
+    left: 100%;
+    transform: translateX(calc(-1 * var(--btn-group-width) + 1px));
+    display: flex;
+    gap: 0px;
+    padding: 0px;
+}
+
+.sort-button {
+    margin: 0;
+    box-sizing: border-box;
+    height: 100%;
+    flex-grow: 1;
+    border-radius: 0;
+    border: 1px solid #aaa;
+    background: #e0e0e0;
+    color: #222;
+    padding: var(--micro);
+}
+
+.sort-button-group > .sort-button:focus-visible {
+    z-index: 2;
+}
+
+.sort-button-group > .sort-button ~ .sort-button {
+    margin-left: -1px;
+}
+
+.sort-button:hover {
+    background: #eeeeee;
+}
+
+.sort-button:active {
+    background: #cccccc;
+}
+
+.sort-button[data-is-active]{
+    background: var(--lightgreen);
+    color: black;
+}

skrub/_reporting/_data/templates/summary-statistics.html

@@ -1,57 +1,100 @@
+{% macro th(name, ascending, descending, is_numeric) %}
+<th class="sort-button-group-wrapper" scope="col">
+    {% if name %}
+    <span class="margin-r-t">{{ name }}</span>
+    {% endif %}
+    <div class="sort-button-group">
+        <button class="sort-button" type="button" data-role="sort-button"
+            data-direction="ascending" title="Sort {{ ascending }}">
+            {% if is_numeric %}
+            {% include "icons/sort-numeric-down.svg" %}
+            {% else %}
+            {% include "icons/sort-alpha-down.svg" %}
+            {% endif %}
+        </button>
+        <button class="sort-button" type="button" data-role="sort-button"
+            data-direction="descending" title="Sort {{ descending }}">
+            {% if is_numeric %}
+            {% include "icons/sort-numeric-down-alt.svg" %}
+            {% else %}
+            {% include "icons/sort-alpha-down-alt.svg" %}
+            {% endif %}
+        </button>
+    </div>
+</th>
+{% endmacro %}
+
+{% macro th1(name, low, high, is_numeric) %}
+{{ th(name, "from columns with " + low + " to columns with " + high, "from columns with " + high + " to columns with " + low, is_numeric) }}
+{% endmacro %}
+
 <article class="wrapper" data-show-on="NON_EMPTY_COLUMN_FILTER_SELECTED"
-         data-hide-on="EMPTY_COLUMN_FILTER_SELECTED">
+    data-hide-on="EMPTY_COLUMN_FILTER_SELECTED">
     <div class="horizontal-scroll">
-        <table class="pure-table pure-table-striped">
+        <table class="pure-table pure-table-bordered summary-stats-table"
+            data-manager="sortableTable">
             <thead>
                 <tr>
-                    <th scope="col">Column name</th>
-                    <th scope="col">dtype</th>
-                    <th scope="col">Null values</th>
-                    <th scope="col">Unique values</th>
-                    <th scope="col">Mean</th>
-                    <th scope="col">Std</th>
-                    <th scope="col">Min</th>
-                    <th scope="col">Median</th>
-                    <th scope="col">Max</th>
+                    {{ th("Column", "from first column to last column", "from last column to first column", True) }}
+                    {{ th("Column name", "by column name from A to Z", "by column name from Z to A", False) }}
+                    {{ th("dtype", "by dtype from A to Z", "by dtype from Z to A", False) }}
+                    {{ th1("Null values", "the fewest null values", "the most null values", True) }}
+                    {{ th1("Unique values", "the fewest unique values", "the most unique values", True) }}
+                    {{ th1("Mean", "the lowest mean", "the highest mean", True) }}
+                    {{ th1("Std", "the lowest standard deviation", "the highest standard deviation", True) }}
+                    {{ th1("Min", "the lowest minimum value", "the highest minimum value", True) }}
+                    {{ th1("Median", "the lowest median", "the highest median", True) }}
+                    {{ th1("Max", "the lowest maximum value", "the highest maximum value", True) }}
                 </tr>
             </thead>
             <tbody>
                 {% for column in summary.columns %}
-                <tr data-manager="FilterableColumn"
-                    data-column-name="{{ column.name }}">
+                <tr data-manager="FilterableColumn" data-column-name="{{ column.name }}"
+                    data-dataframe-column-idx="{{ loop.index0 }}">
+                    <td data-value="{{ loop.index0 }}" data-numeric>{{ loop.index0 }}
+                    </td>
                     <td>{{ column.name }}</td>
                     <td>{{ column.dtype }}</td>
-                    <td class="{{ column.nulls_level }}">
+                    <td class="{{ column.nulls_level }}"
+                        data-value="{{ column.null_count }}" data-numeric>
                         {{ column.null_count }} ({{ column.null_proportion | format_percent }})
                     </td>
 
                     {% if column.n_unique %}
-                    <td>{{ column.n_unique }} ({{ column.unique_proportion | format_percent }})
+                    <td data-value="{{ column.n_unique }}" data-numeric>
+                        {{ column.n_unique }} ({{ column.unique_proportion | format_percent }})
                     </td>
                     {% else %}
-                    <td></td>
+                    <td data-value="nan" data-numeric></td>
                     {% endif %}
 
                     {% if "mean" in column %}
-                    <td>{{ column["mean"] | format_number }}</td>
-                    <td>{{ column["standard_deviation"] | format_number }}</td>
+                    <td data-value="{{ column['mean'] }}" data-numeric>
+                        {{ column["mean"] | format_number }}</td>
+                    <td data-value="{{ column['standard_deviation'] }}" data-numeric>
+                        {{ column["standard_deviation"] | format_number }}</td>
                     {% else %}
-                    <td></td>
-                    <td></td>
+                    <td data-value="nan" data-numeric></td>
+                    <td data-value="nan" data-numeric></td>
                     {% endif %}
 
                     {% if column.quantiles %}
-                    <td>{{ column.quantiles[0.0] | format_number }}</td>
-                    <td>{{ column.quantiles[0.5] | format_number }}</td>
-                    <td>{{ column.quantiles[1.0] | format_number }}</td>
+                    <td data-value="{{ column.quantiles[0.0] }}" data-numeric>
+                        {{ column.quantiles[0.0] | format_number }}</td>
+                    <td data-value="{{ column.quantiles[0.5] }}" data-numeric>
+                        {{ column.quantiles[0.5] | format_number }}</td>
+                    <td data-value="{{ column.quantiles[1.0] }}" data-numeric>
+                        {{ column.quantiles[1.0] | format_number }}</td>
                     {% elif "min" in column %}
-                    <td>{{ column.min | format_number }}</td>
-                    <td></td>
-                    <td>{{ column.max | format_number }}</td>
+                    <td data-value="{{ column.min }}" data-numeric>
+                        {{ column.min | format_number }}</td>
+                    <td data-value="nan" data-numeric></td>
+                    <td data-value="{{ column.max }}" data-numeric>
+                        {{ column.max | format_number }}</td>
                     {% else %}
-                    <td></td>
-                    <td></td>
-                    <td></td>
+                    <td data-value="nan" data-numeric></td>
+                    <td data-value="nan" data-numeric></td>
+                    <td data-value="nan" data-numeric></td>
                     {% endif %}
 
                 </tr>

skrub/_reporting/js_tests/cypress/e2e/summary-statistics.cy.js

@@ -0,0 +1,29 @@
+describe('test sorting the summary stats columns', () => {
+    it('sorts the table when clicking arrows', () => {
+        cy.get('@report').find('[data-test="summary-statistics-tab"]')
+            .click();
+        cy.get('@report').find('.summary-stats-table').as('table');
+        cy.get('@table').find('tbody tr').first().should('have.attr',
+            'data-column-name', 'gender');
+        cy.get('@report').contains('Column name').as('colName');
+        cy.get('@colName').parent().find('button').first().as('colNameButton').click();
+        cy.get('@colNameButton').should('have.attr', 'data-is-active');
+        cy.get('@table').find('tbody tr').first().should('have.attr',
+                                                         'data-column-name', 'assignment_category');
+        cy.get('@report').find('th').contains('Unique values').as(
+            'unique');
+        cy.get('@unique').parent().find('button').first().as('uniqueButton').click();
+        cy.get('@uniqueButton').should('have.attr', 'data-is-active');
+        cy.get('@colNameButton').should('not.have.attr', 'data-is-active');
+        cy.get('@table').find('tbody tr').first().should('have.attr',
+            'data-column-name', 'gender');
+        cy.get('@table').find('tbody tr').last().should('have.attr',
+            'data-column-name', 'year_first_hired');
+        cy.get('@unique').parent().find('button').first().next()
+    .click();
+        cy.get('@table').find('tbody tr').first().should('have.attr',
+            'data-column-name', 'date_first_hired');
+        cy.get('@table').find('tbody tr').last().should('have.attr',
+            'data-column-name', 'year_first_hired');
+    });
+});