Predict with JavaScript

1. Install.

dependencies: { "@tangramdotdev/tangram": "*", }

2. Predict.

First, import the tangram library and load the model file. Then, make an object with info for a new patient that matches the CSV, excluding the diagnosis column. Finally, call predict and print out the result.

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
const fs = require("fs"); const path = require("path"); const tangram = require("@tangramdotdev/tangram"); // Get the path to the .tangram file. const modelPath = path.join(__dirname, "heart_disease.tangram"); // Load the model from the path. const modelData = fs.readFileSync(modelPath); const model = new tangram.Model(modelData.buffer); // Create an example input matching the schema of the CSV file the model was trained on. // Here the data is just hard-coded, but in your application you will probably get this // from a database or user input. const input = { age: 63, gender: "male", chest_pain: "typical angina", resting_blood_pressure: 145, cholesterol: 233, fasting_blood_sugar_greater_than_120: "true", resting_ecg_result: "probable or definite left ventricular hypertrophy", exercise_max_heart_rate: 150, exercise_induced_angina: "no", exercise_st_depression: 2.3, exercise_st_slope: "downsloping", fluoroscopy_vessels_colored: "0", thallium_stress_test: "fixed defect", }; // Make the prediction! const output = model.predict(input); // Print the output. console.log("Output:", output);