nextbrain
v1.4.0
Published
Convenient access to the NextBrain API from javascript
Downloads
52
Maintainers
Readme
NextBrain AI
Convenient access to the NextBrain AI API from Javascript
Installation
npm install nextbrain
All steps in one.
import { NextBrain } from 'nextbrain'
const nb = new NextBrain({
access_token: '<YOUR-ACCESS-TOKEN-HERE>',
})
// You can create your custom table and predict table by your own from any source
// It is a list of list, where the first row contains the header
// Example:
// [
// [ Column1, Column2, Column3 ],
// [ 1, 2, 3 ],
// [ 4, 5, 6 ]
// ]
const table = await nb.loadCsv('<PATH-TO-YOUR-TRAINING-CSV>')
const predictTable = await nb.loadCsv('<PATH-TO-YOUR-PREDICTING-CSV>')
const [modelId, response] = await nb.uploadAndPredict(table, predictTable, '<YOUR-TARGET-COLUMN>')
console.log('Response:', response)
// You can optionally delete the model
await nb.deleteModel(modelId)
Step by step
import { NextBrain } from 'nextbrain'
const nb = new NextBrain({
access_token: '<YOUR-ACCESS-TOKEN-HERE>',
})
// You can create your custom table and predict table by your own from any source
const table = await nb.loadCsv('<PATH-TO-YOUR-TRAINING-CSV>')
// Upload the model to NextBrain service
const modelId = await nb.uploadModel(table)
// Train the model
// You can re-train a previous model
await nb.trainModel(modelId, '<YOUR-TARGET-COLUMN>')
const predictTable = await nb.loadCsv('<PATH-TO-YOUR-PREDICTING-CSV>')
// You can predict multiple using the same model (don't need to create a new model each time)
const response = await nb.predictModel(modelId, predictTable)
console.log('Response:', response)
Extra notes
Everytime you train, you can select an option to create lightning models. isLightning
is an optional parameter that by default is set to false
but can be overrided in trainModel
and uploadAndPredict
.
We also recommend that you investigate all the methods that the class provides you with to make the most of the functionalities we offer. For example, you can use the getAccuracy
method to obtain all the information about the performance of your model.