yggdrasil-decision-forests

YDF in NPM

This package makes it possible to run decision forest models in browser and with NodeJS.

Usage example

The first step is to train a decision forests model using YDF.

Let's train a model in Python.

import ydf
import pandas as pd

# Download a training dataset
ds_path = "https://raw.githubusercontent.com/google/yggdrasil-decision-forests/main/yggdrasil_decision_forests/test_data/dataset/"
train_ds = pd.read_csv(ds_path + "adult_train.csv")

# Train a Gradient Boosted Trees model
model = ydf.GradientBoostedTreesLearner(label="income", pure_serving_model=True).train(train_ds)

# Save the model
model.save("/tmp/my_model")

# DO NOT SUBMIT => ZIP or Serialized.

With NodeJS and CommonJS

(async function (){
    // Load the YDF library
    const ydf = await require("yggdrasil-decision-forests")();

    // Load the model
    const fs = require("node:fs");
    let model = await ydf.loadModelFromZipBlob(fs.readFileSync("./model.zip"));

    // Create a batch of examples.
    let examples = {
        "f1": [2, 6, null, 3],
        "f2": ["A", "A", "B", "A"],
    };


    // Make predictions
    let predictions = model.predict(examples);
    console.log("predictions:", predictions);

    // Release model
    model.unload();
}())

With NodeJS and ES6

import * as fs from "node:fs";
import YggdrasilDecisionForests from 'yggdrasil-decision-forests';

// Load the YDF library
let ydf = await YggdrasilDecisionForests();

// Load the model
let model = await ydf.loadModelFromZipBlob(fs.readFileSync("./model.zip"));

// Create a batch of examples.
let examples = {
        "f1": [2, 6, null, 3],
        "f2": ["A", "A", "B", "A"],
    };

// Make predictions
let predictions = model.predict(examples);
console.log("predictions:", predictions);

// Release model
model.unload();

In Browser

<script src="./node_modules/yggdrasil-decision-forests/dist/inference.js"></script>
<script src="https://cdnjs.cloudflare.com/ajax/libs/jszip/3.10.0/jszip.min.js"></script>
<script>
YggdrasilDecisionForests()
    .then(ydf => ydf.loadModelFromUrl("http://localhost:3000/model.zip"))
    .then(model => {
        let examples = {
            "f1": [2, 6, null, 3],
            "f2": ["A", "A", "B", "A"],
        };
        predictions = model.predict(examples);
        model.unload();
    });
</script>

Developer

Run unit tests

npm test

Update the binary bundle

# Download YDF
git clone google/yggdrasil-decision-forests

# Compile the YDF with WebAssembly
yggdrasil_decision_forests/port/javascript/tools/build_zipped_library.sh

# Extract the the content of `yggdrasil_decision_forests/dist/ydf.zip` in `dist`.