85 lines
3.6 KiB
Java
85 lines
3.6 KiB
Java
"use strict";
|
|
Object.defineProperty(exports, "__esModule", { value: true });
|
|
var tslib_1 = require("tslib");
|
|
/*
|
|
* kernel density estimation
|
|
*/
|
|
var util_1 = require("@antv/util");
|
|
var get_series_values_1 = tslib_1.__importDefault(require("../util/get-series-values"));
|
|
var kernel_1 = tslib_1.__importDefault(require("../util/kernel"));
|
|
var bw = tslib_1.__importStar(require("../util/bandwidth"));
|
|
var partition_1 = require("../util/partition");
|
|
var simple_statistics_1 = require("simple-statistics");
|
|
var helper_1 = require("../helper");
|
|
var DEFAULT_OPTIONS = {
|
|
minSize: 0.01,
|
|
as: ['key', 'y', 'size'],
|
|
// fields: [ 'y1', 'y2' ], // required, one or more fields
|
|
// extent: [], // extent to execute regression function, default: [ [ min(x), max(x) ], [ min(y), max(y) ] ]
|
|
method: 'gaussian',
|
|
bandwidth: 'nrd',
|
|
};
|
|
var KERNEL_METHODS = util_1.keys(kernel_1.default);
|
|
function kde(rows, options) {
|
|
var opts = helper_1.mergeOptions(options, DEFAULT_OPTIONS);
|
|
var fields = options.fields, groupBy = options.groupBy, bandwidth = options.bandwidth, method = options.method;
|
|
if (!fields || !Array.isArray(fields) || fields.length < 1) {
|
|
throw new TypeError('invalid fields: must be an array of at least 1 strings!');
|
|
}
|
|
if (!opts.as || !Array.isArray(opts.as) || opts.as.length !== 3) {
|
|
throw new TypeError('invalid as: must be an array of 3 strings!');
|
|
}
|
|
var methodFunc;
|
|
if (typeof method === 'string') {
|
|
if (KERNEL_METHODS.indexOf(method) === -1) {
|
|
throw new TypeError("invalid method: " + method + ". Must be one of " + KERNEL_METHODS.join(', '));
|
|
}
|
|
methodFunc = kernel_1.default[method];
|
|
}
|
|
if (util_1.isFunction(method)) {
|
|
methodFunc = method;
|
|
}
|
|
var extent = opts.extent;
|
|
if (!extent || !Array.isArray(extent) || extent.length === 0) {
|
|
var rangeArr_1 = [];
|
|
fields.forEach(function (field) {
|
|
rangeArr_1.push.apply(rangeArr_1, tslib_1.__spread(helper_1.range(rows.map(function (item) { return item[field]; }))));
|
|
});
|
|
extent = helper_1.range(rangeArr_1);
|
|
}
|
|
var step = 0;
|
|
if (typeof bandwidth === 'string' && bw[bandwidth]) {
|
|
step = bw[bandwidth](rows.map(function (item) { return item[fields[0]]; }));
|
|
}
|
|
else if (util_1.isFunction(bandwidth)) {
|
|
step = bandwidth(rows.map(function (item) { return item[fields[0]]; }));
|
|
}
|
|
else if (!util_1.isNumber(bandwidth) || bandwidth <= 0) {
|
|
step = bw.nrd(rows.map(function (item) { return item[fields[0]]; }));
|
|
}
|
|
var seriesValues = get_series_values_1.default(extent, step);
|
|
var result = [];
|
|
var groups = partition_1.partition(rows, groupBy);
|
|
util_1.forIn(groups, function (group) {
|
|
var probalityDensityFunctionByField = {};
|
|
util_1.each(fields, function (field) {
|
|
var row = util_1.pick(group[0], groupBy);
|
|
probalityDensityFunctionByField[field] = simple_statistics_1.kernelDensityEstimation(group.map(function (item) { return item[field]; }), methodFunc, step);
|
|
var _a = tslib_1.__read(opts.as, 3), key = _a[0], y = _a[1], size = _a[2];
|
|
row[key] = field;
|
|
row[y] = [];
|
|
row[size] = [];
|
|
util_1.each(seriesValues, function (yValue) {
|
|
var sizeValue = probalityDensityFunctionByField[field](yValue);
|
|
if (sizeValue >= opts.minSize) {
|
|
row[y].push(yValue);
|
|
row[size].push(sizeValue);
|
|
}
|
|
});
|
|
result.push(row);
|
|
});
|
|
});
|
|
return result;
|
|
}
|
|
exports.kde = kde;
|