davidkyle

time,airline,responsetime,sourcetype
1403481600000,AAL,132.2046,farequote
1403481600000,JZA,990.4628,farequote
1403481600000,JBU,877.5927,farequote
1403481600000,KLM,1355.4812,farequote
1403481600000,NKS,9991.3981,farequote
1403481600000,TRS,412.1948,farequote
1403481600000,DAL,401.4948,farequote
1403481600000,FFT,181.5529,farequote
1403481600000,SWA,160.214,farequote

davidkyle

#!/bin/sh


curl -XPUT -u 'elastic:password' -H 'Content-Type: application/json' 'http://localhost:9200/_ml/inference/density_model?pretty' -d '
{
  "input": {
    "field_names": [
      "key_metric"]
  },
  "description": "test model for regression",

davidkyle

from typing import List


def add_if_exists(d: dict, k: str, v) -> dict:
    """
    :param v:
    :param k:
    :type d: object
    """
    if v is not None:

davidkyle

{
  "size": 1,
  "aggs": {
    "session": {
      "composite": {
        "size": 10,
        "sources": [
          {
            "session": {
              "terms": {

davidkyle

Getting started with the Inference Aggregation

First ingest the customer churn data and trained model

• calls.csv
• customers.csv

Ingest the linked csv files above into Elastic using the Kibana Data Visualizer. You will find the Data Visualizer in the Machine Learning section. Ingest into the indices calls and customers.

• telco_churn_model.json

davidkyle

from elasticsearch import Elasticsearch

from eland.ml.pytorch import PyTorchModel
from eland.ml.pytorch.nlp_ml_model import (
    TextExpansionInferenceOptions, 
    NlpTrainedModelConfig, 
    TrainedModelInput, 
    NlpBertTokenizationConfig
)