A tree mapping of a BigML decision tree built on the iris dataset.
The outer most rectangle represents the root of the tree. Each inner retangle represents the children of the outer rectangle's nodes. The number of training instances captured by a node determine its size.
| <!DOCTYPE html> | |
| <meta charset="utf-8"> | |
| <style> | |
| body { | |
| font-family: "Helvetica Neue", Helvetica, Arial, sans-serif; | |
| margin: auto; | |
| position: relative; | |
| width: 960px; | |
| background: #fff; | |
| } | |
| #color-controls { | |
| font: 14px sans-serif; | |
| position: absolute; | |
| right: 10px; | |
| top: 10px; | |
| padding: 3px; | |
| } | |
| #color-controls div { | |
| padding: 4px; | |
| } | |
| #hover-info { | |
| font: 14px sans-serif; | |
| position: absolute; | |
| left: 10px; | |
| top: 10px; | |
| } | |
| #summary-info { | |
| font: 14px sans-serif; | |
| position: absolute; | |
| left: 10px; | |
| bottom: 10px; | |
| font: 12px sans-serif; | |
| } | |
| #summary-info div { | |
| padding: 2px; | |
| } | |
| .split-predicate { | |
| font-weight:bold; | |
| border-bottom: 1px solid #DFDFDF; | |
| padding: 7px; | |
| } | |
| .node-info { | |
| margin-top: 10px; | |
| } | |
| .node-info td { | |
| padding: 2px 7px 2px; | |
| } | |
| </style> | |
| <body> | |
| <div id="color-controls"> | |
| <form> | |
| <div> | |
| <input type="radio" name="mode" value="prediction" checked \> | |
| <label>Prediction</label> | |
| </div> | |
| <div> | |
| <input type="radio" name="mode" value="confidence"\> | |
| <label id="cnf">Confidence</label> | |
| </div> | |
| <div> | |
| <input type="radio" name="mode" value="split"\> | |
| <label>Split Field</label> | |
| </div> | |
| </form> | |
| </div> | |
| <div id="hover-info"></div> | |
| <div id="summary-info"></div> | |
| </body> | |
| <script src="http://d3js.org/d3.v3.min.js"></script> | |
| <script> | |
| var width = 960, | |
| height = 600; | |
| function hover_adjust(d, color) { | |
| return d.hover ? d3.rgb(color).brighter(0.66) : color; | |
| } | |
| var treemap = d3.layout.treemap() | |
| .padding(10) | |
| .size([width - 50, height - 190]) | |
| .sticky(true) | |
| .value(function(d) { return d.count; }); | |
| var svg = d3.select("body").append("svg:svg") | |
| .attr("width", width) | |
| .attr("height", height) | |
| .append("svg:g") | |
| .attr("transform", "translate(20,120)"); | |
| function find_minmax(node, attr) { | |
| if (node.children) { | |
| minmaxs = node.children.map(function (n) { return find_minmax(n, attr); }); | |
| min = Math.min.apply(null, minmaxs.map(function (mm) {return mm.min})); | |
| max = Math.max.apply(null, minmaxs.map(function (mm) {return mm.max})); | |
| return {"min": Math.min(min, node[attr]), "max": Math.max(max, node[attr])}; | |
| } else { | |
| return {"min": node[attr], "max": node[attr]}; | |
| } | |
| } | |
| d3.json("iris-model.json", function(error, root) { | |
| var model = root.model; | |
| var model_type = model.model_fields[root.objective_field].optype == "categorical" ? | |
| "classification" : "regression"; | |
| var minmaxs = {}; | |
| if (model_type == "classification") { | |
| minmaxs.confidence = find_minmax(model.root, "confidence"); | |
| } else { | |
| // Hacky label switch for regression trees | |
| document.getElementById("cnf").innerHTML = "Expected Error"; | |
| minmaxs.expected_error = find_minmax(model.root, "confidence"); | |
| minmaxs.output = find_minmax(model.root, "output"); | |
| } | |
| var scale_pred = model_type == "classification" ? | |
| d3.scale.category10() : | |
| d3.scale.linear().domain([minmaxs.output.min, | |
| minmaxs.output.max]) | |
| .range(["#222", "#2ee"]); | |
| var scale_conf = model_type == "classification" ? | |
| d3.scale.linear().domain([minmaxs.confidence.min, | |
| minmaxs.confidence.max]) | |
| .range(["#d33", "#3d3"]) : | |
| d3.scale.linear().domain([minmaxs.expected_error.max, | |
| minmaxs.expected_error.min]) | |
| .range(["#d33", "#3d3"]); | |
| var scale_split = d3.scale.category20b(); | |
| var color_lookup = | |
| {"prediction": function(d) { return hover_adjust(d, scale_pred(d.output)); }, | |
| "confidence": function(d) { return hover_adjust(d, scale_conf(d.confidence)); }, | |
| "split": function(d) { return hover_adjust(d, scale_split(d.predicate.field)); } }; | |
| var color_fn = color_lookup["prediction"]; | |
| var path = svg.selectAll("g") | |
| .data(treemap.nodes(root.model.root)) | |
| .enter().append("svg:g") | |
| .attr("class", "cell") | |
| .attr("transform", function(d) { return "translate(" + d.x + "," + d.y + ")"; }) | |
| .append("svg:rect") | |
| .attr("width", function(d) { return d.dx; }) | |
| .attr("height", function(d) { return d.dy; }) | |
| .style("fill", color_fn) | |
| .style("stroke", "#333") | |
| .on("mouseover", mouseover) | |
| .on("mouseout", mouseout); | |
| d3.selectAll("input").on("change", change); | |
| function change() { | |
| color_fn = color_lookup[this.value]; | |
| path.transition().duration(250).style("fill", color_fn); | |
| } | |
| function mouseover(d) { | |
| var split = d.predicate; | |
| var split_msg = split.field ? model.model_fields[split.field].name + " " + | |
| split.operator + " " + split.value | |
| : "Tree Root"; | |
| var conf_msg = {"classification": "Confidence", "regression": "Expected Error"}; | |
| var hover = d3.select("#hover-info"); | |
| hover.append("div").attr("class", "split-predicate").text(split_msg); | |
| tbody = hover.append("table").attr("class", "node-info").append("tbody"); | |
| var output = model_type == "classification" ? d.output : parseFloat(d.output.toFixed(3)); | |
| table_add(tbody, "Prediction", output); | |
| table_add(tbody, conf_msg[model_type], parseFloat(d.confidence.toFixed(3))); | |
| table_add(tbody, "Count", d.count); | |
| mark_hover(d, true); | |
| path.style("fill", color_fn); | |
| var summ_doc = d3.select("#summary-info"); | |
| var summaries = summarize(d); | |
| for (id in summaries) { | |
| if (!summaries.hasOwnProperty(id)) { continue; } | |
| var name = model.model_fields[id].name; | |
| var msg = name; | |
| fs = summaries[id]; | |
| if (isNum(fs.min)) { | |
| msg = parseFloat(fs.min.toFixed(3)) + " < " + msg; | |
| } | |
| if (isNum(fs.max)) { | |
| msg += " <= " + parseFloat(fs.max.toFixed(3)); | |
| } | |
| if (fs.eq) { | |
| msg += " = " + fs.eq; | |
| } else if (fs.not_eq) { | |
| msg += " !="; | |
| var first = true; | |
| for (category in fs.not_eq) { | |
| if (first) { | |
| first = false; | |
| } else { | |
| msg += "|"; | |
| } | |
| if (!fs.not_eq.hasOwnProperty(category)) { continue; } | |
| msg += " " + category; | |
| } | |
| msg = msg; | |
| } | |
| summ_doc.append("div").text(msg); | |
| } | |
| } | |
| function mouseout(d) { | |
| d3.select("#hover-info").html(""); | |
| d3.select("#summary-info").html(""); | |
| mark_hover(d, false); | |
| path.style("fill", color_fn); | |
| } | |
| function mark_hover (d, val) { | |
| if (d.parent) { mark_hover(d.parent, val); }; | |
| d.hover = val; | |
| } | |
| }); | |
| d3.select(self.frameElement).style("height", height + "px"); | |
| function isNum(n) { | |
| return !isNaN(parseFloat(n)) && isFinite(n); | |
| } | |
| function summarize (node) { | |
| var pred = node.predicate; | |
| if (node.parent) { | |
| var summary = summarize(node.parent); | |
| switch(pred.operator) { | |
| case "<=": | |
| if (summary[pred.field]) { | |
| var old_max = summary[pred.field].max; | |
| max = isNum(old_max) ? Math.min(pred.value, old_max) : pred.value; | |
| summary[pred.field].max = max; | |
| } else { | |
| summary[pred.field] = {"max": pred.value}; | |
| } | |
| break; | |
| case ">": | |
| if (summary[pred.field]) { | |
| var old_min = summary[pred.field].min; | |
| min = isNum(old_min) ? Math.max(pred.value, old_min) : pred.value; | |
| summary[pred.field].min = min; | |
| } else { | |
| summary[pred.field] = {"min": pred.value}; | |
| } | |
| break; | |
| case "=": | |
| summary[pred.field] = {"eq": pred.value}; | |
| break; | |
| case "!=": | |
| if (!summary[pred.field]) { | |
| summary[pred.field] = {}; | |
| } | |
| if (!summary[pred.field].not_eq) { | |
| summary[pred.field].not_eq = {}; | |
| } | |
| summary[pred.field].not_eq[pred.value] = true; | |
| break; | |
| } | |
| return summary; | |
| } else { | |
| return {}; | |
| } | |
| } | |
| function table_add (table, field, val) { | |
| var row = table.append("tr"); | |
| row.append("td").text(field); | |
| row.append("td").text(val); | |
| return row; | |
| } | |
| </script> |
| {"public_dataset": false, "code": 200, "locale": "en-US", "short_url": "", "image": "gallery/model.png", "job_type": 400, "private": true, "dataset": "dataset/514bd86f035d07049200003f", "fields_meta": {"count": 5, "total": 5, "limit": 1000, "offset": 0}, "white_box": false, "dataset_id": "514bd86f035d07049200003f", "randomize": false, "id": "514fb3d00c0b5e04c3000093", "number_of_predictions": 0, "category": 0, "modified_fields": {}, "rows": 150, "out_of_bag": false, "ordering": 0, "fields_to_show": [{"optype": "numeric", "is_objective": false, "name": "sepal length", "datatype_name": "Numeric", "tooltip": "Numeric", "stats": [["Minimum", "4.30"], ["Mean", "5.84"], ["Median", "5.78"], ["Maximum", "7.90"], ["Std dev", "0.83"]], "id": "000000", "css": "numeric type_label datatype-tooltip"}, {"optype": "numeric", "is_objective": false, "name": "sepal width", "datatype_name": "Numeric", "tooltip": "Numeric", "stats": [["Minimum", "2.00"], ["Mean", "3.06"], ["Median", "3.02"], ["Maximum", "4.40"], ["Std dev", "0.44"]], "id": "000001", "css": "numeric type_label datatype-tooltip"}, {"optype": "numeric", "is_objective": false, "name": "petal length", "datatype_name": "Numeric", "tooltip": "Numeric", "stats": [["Minimum", "1.00"], ["Mean", "3.76"], ["Median", "4.34"], ["Maximum", "6.90"], ["Std dev", "1.77"]], "id": "000002", "css": "numeric type_label datatype-tooltip"}, {"optype": "numeric", "is_objective": false, "name": "petal width", "datatype_name": "Numeric", "tooltip": "Numeric", "stats": [["Minimum", "0.10"], ["Mean", "1.20"], ["Median", "1.33"], ["Maximum", "2.50"], ["Std dev", "0.76"]], "id": "000003", "css": "numeric type_label datatype-tooltip"}, {"optype": "categorical", "is_objective": true, "name": "species", "datatype_name": "Categorical", "tooltip": "Categorical", "id": "000004", "css": "categorical type_label datatype-tooltip"}], "views": 0, "size": 4608, "range": [1, 150], "credits_per_prediction": 0.0, "source": "source/514bcf9d0c0b5e3fd100026b", "number_of_public_predictions": 0, "sample_rate": 1.0, "objective_fields": ["000004"], "user_name": "ashenfad", "columns": 5, "selective_pruning": false, "status": {"progress": 1.0, "message": "The model has been created", "code": 5, "elapsed": 89}, "updated": "2013-03-25T02:17:52.927000", "description": "", "tags": [], "price": 0.0, "excluded_fields": [], "cloned": false, "credits": 0.017578125, "stat_pruning": true, "objective_field": "000004", "clones": 0, "resource": "model/514fb3d00c0b5e04c3000093", "name": "Iris Model", "created": "Mon, 25 Mar 2013 02:17:52 +0000", "url": "", "dataset_status": true, "source_status": true, "number_of_evaluations": 0, "max_columns": 5, "max_rows": 150, "input_fields": ["000000", "000001", "000002", "000003"], "published": null, "source_id": "514bcf9d0c0b5e3fd100026b", "model": {"kind": "stree", "missing_strategy": "Last prediction", "importance": [["000002", 0.53556], ["000003", 0.46444], ["000000", 0], ["000001", 0]], "fields": {"000004": {"optype": "categorical", "name": "species", "datatype": "string", "preferred": true, "summary": {"missing_count": 0, "categories": [["Iris-versicolor", 50], ["Iris-setosa", 50], ["Iris-virginica", 50]]}, "column_number": 4, "order": 4}, "000002": {"optype": "numeric", "name": "petal length", "datatype": "double", "preferred": true, "summary": {"sum_squares": 2582.71, "splits": [1.25138, 1.32426, 1.37171, 1.40962, 1.44567, 1.48173, 1.51859, 1.56301, 1.6255, 1.74645, 3.23033, 3.675, 3.94203, 4.0469, 4.18243, 4.34142, 4.45309, 4.51823, 4.61771, 4.72566, 4.83445, 4.93363, 5.03807, 5.1064, 5.20938, 5.43979, 5.5744, 5.6646, 5.81496, 6.02913, 6.38125], "missing_count": 0, "sum": 563.7, "median": 4.34142, "maximum": 6.9, "minimum": 1, "standard_deviation": 1.7653, "variance": 3.11628, "population": 150, "bins": [[1, 1], [1.1, 1], [1.2, 2], [1.3, 7], [1.4, 13], [1.5, 13], [1.63636, 11], [1.9, 2], [3, 1], [3.3, 2], [3.5, 2], [3.6, 1], [3.75, 2], [3.9, 3], [4.0375, 8], [4.23333, 6], [4.46667, 12], [4.6, 3], [4.74444, 9], [4.94444, 9], [5.1, 8], [5.25, 4], [5.46, 5], [5.6, 6], [5.75, 6], [5.95, 4], [6.1, 3], [6.3, 1], [6.4, 1], [6.6, 1], [6.7, 2], [6.9, 1]], "mean": 3.758}, "column_number": 2, "order": 2}, "000003": {"optype": "numeric", "name": "petal width", "datatype": "double", "preferred": true, "summary": {"sum_squares": 302.33, "missing_count": 0, "sum": 179.9, "median": 1.32848, "maximum": 2.5, "minimum": 0.1, "standard_deviation": 0.76224, "variance": 0.58101, "counts": [[0.1, 5], [0.2, 29], [0.3, 7], [0.4, 7], [0.5, 1], [0.6, 1], [1, 7], [1.1, 3], [1.2, 5], [1.3, 13], [1.4, 8], [1.5, 12], [1.6, 4], [1.7, 2], [1.8, 12], [1.9, 5], [2, 6], [2.1, 6], [2.2, 3], [2.3, 8], [2.4, 3], [2.5, 3]], "population": 150, "mean": 1.19933}, "column_number": 3, "order": 3}, "000000": {"optype": "numeric", "name": "sepal length", "datatype": "double", "preferred": true, "summary": {"sum_squares": 5223.85, "splits": [4.51526, 4.67252, 4.81113, 4.89582, 4.96139, 5.01131, 5.05992, 5.11148, 5.18177, 5.35681, 5.44129, 5.5108, 5.58255, 5.65532, 5.71658, 5.77889, 5.85381, 5.97078, 6.05104, 6.13074, 6.23023, 6.29578, 6.35078, 6.41459, 6.49383, 6.63013, 6.70719, 6.79218, 6.92597, 7.20423, 7.64746], "missing_count": 0, "sum": 876.5, "median": 5.77889, "maximum": 7.9, "minimum": 4.3, "standard_deviation": 0.82807, "variance": 0.68569, "population": 150, "bins": [[4.3, 1], [4.425, 4], [4.6, 4], [4.7, 2], [4.8, 5], [4.9, 6], [5, 10], [5.1, 9], [5.2, 4], [5.3, 1], [5.4, 6], [5.5, 7], [5.6, 6], [5.7, 8], [5.8, 7], [5.9, 3], [6, 6], [6.1, 6], [6.2, 4], [6.3, 9], [6.44167, 12], [6.6, 2], [6.7, 8], [6.8, 3], [6.92, 5], [7.1, 1], [7.2, 3], [7.3, 1], [7.4, 1], [7.6, 1], [7.7, 4], [7.9, 1]], "mean": 5.84333}, "column_number": 0, "order": 0}, "000001": {"optype": "numeric", "name": "sepal width", "datatype": "double", "preferred": true, "summary": {"sum_squares": 1430.4, "missing_count": 0, "sum": 458.6, "median": 3.02044, "maximum": 4.4, "minimum": 2, "standard_deviation": 0.43587, "variance": 0.18998, "counts": [[2, 1], [2.2, 3], [2.3, 4], [2.4, 3], [2.5, 8], [2.6, 5], [2.7, 9], [2.8, 14], [2.9, 10], [3, 26], [3.1, 11], [3.2, 13], [3.3, 6], [3.4, 12], [3.5, 6], [3.6, 4], [3.7, 3], [3.8, 6], [3.9, 2], [4, 1], [4.1, 1], [4.2, 1], [4.4, 1]], "population": 150, "mean": 3.05733}, "column_number": 1, "order": 1}}, "model_fields": {"000004": {"optype": "categorical", "name": "species", "datatype": "string", "preferred": true, "column_number": 4, "order": 4}, "000002": {"optype": "numeric", "name": "petal length", "datatype": "double", "preferred": true, "column_number": 2, "order": 2}, "000003": {"optype": "numeric", "name": "petal width", "datatype": "double", "preferred": true, "column_number": 3, "order": 3}}, "support_threshold": 0, "split_criterion": "Information gain mix", "root": {"count": 150, "confidence": 0.26289, "predicate": true, "objective_summary": {"categories": [["Iris-versicolor", 50], ["Iris-setosa", 50], ["Iris-virginica", 50]]}, "output": "Iris-virginica", "children": [{"count": 50, "output": "Iris-setosa", "confidence": 0.92865, "predicate": {"operator": "<=", "field": "000002", "value": 2.45}, "objective_summary": {"categories": [["Iris-setosa", 50]]}}, {"count": 100, "confidence": 0.40383, "predicate": {"operator": ">", "field": "000002", "value": 2.45}, "objective_summary": {"categories": [["Iris-versicolor", 50], ["Iris-virginica", 50]]}, "output": "Iris-virginica", "children": [{"count": 48, "output": "Iris-virginica", "confidence": 0.86024, "predicate": {"operator": ">", "field": "000003", "value": 1.65}, "objective_summary": {"categories": [["Iris-versicolor", 2], ["Iris-virginica", 46]]}}, {"count": 52, "confidence": 0.81826, "predicate": {"operator": "<=", "field": "000003", "value": 1.65}, "objective_summary": {"categories": [["Iris-virginica", 4], ["Iris-versicolor", 48]]}, "output": "Iris-versicolor", "children": [{"count": 47, "output": "Iris-versicolor", "confidence": 0.92444, "predicate": {"operator": "<=", "field": "000002", "value": 4.95}, "objective_summary": {"categories": [["Iris-versicolor", 47]]}}, {"count": 5, "output": "Iris-virginica", "confidence": 0.37553, "predicate": {"operator": ">", "field": "000002", "value": 4.95}, "objective_summary": {"categories": [["Iris-versicolor", 1], ["Iris-virginica", 4]]}}]}]}]}, "depth_threshold": 20}, "replacement": false} |
