kylegallatin

from sklearn.pipeline import make_pipeline
from sklearn.linear_model import LogisticRegression
from sklearn.preprocessing import StandardScaler
from sklearn.model_selection import train_test_split

## logistic regression parameter config
parameters = {
    "penalty":"l2",
    "C":1.0,
    "max_iter": 100

kylegallatin

import csv
import joblib
import sklearn.pipeline
from datetime import datetime

## helper function to save a model/pipeline
def save_pipeline(pipeline: sklearn.pipeline.Pipeline, ts: datetime = None) -> str:
    model_path = f"data/pipeline-{ts}.pkl".replace(" ","-")
    joblib.dump(pipeline, model_path)
    return model_path

kylegallatin

from typing import Iterable, Callable, Dict

class SuperSimpleFeatureStore:
    
    def __init__(self, dataframe: pd.DataFrame):
        self.dataframe = dataframe
        self.feature_dict = dataframe.set_index("user_id").T.to_dict()
            
    def register_feature(self, name: str, feature_definition: Callable) -> None:
        for key in self.feature_dict:

kylegallatin

import pandas as pd
from collections import Counter

print("LinkedIn Data Summary", "-"*50, sep = "\n")
connections = pd.read_csv("Connections.csv", parse_dates = ["Connected On"], skiprows=2)["Connected On"]
print("Total connections:", len(connections))
d = Counter(connections.apply(lambda x: x.year))
print("\n".join([f"{v} new connections in {k}" for k,v in d.items()][::-1]), "so far 😉", "\n", "-"*50)

messages = pd.read_csv("messages.csv", parse_dates = ["DATE"])

kylegallatin

apiVersion: networking.k8s.io/v1
kind: Ingress
metadata:
  name: ml-app-ingress
  annotations:
    nginx.ingress.kubernetes.io/rewrite-target: /$1
spec:
  rules:
    - host: localhost
      http:

kylegallatin

apiVersion: v1
kind: Service
metadata:
  name: ml-app-service
spec:
  selector:
    app: ml-app
  ports:
    - protocol: TCP
      port: 5000

kylegallatin

apiVersion: apps/v1
kind: Deployment
metadata:
  name: ml-app-deployment
  labels:
    app: ml-app
spec:
  replicas: 1
  selector:
    matchLabels:

kylegallatin

dash==1.20.0
dash-bootstrap-components==0.12.2
dash-core-components==1.16.0
dash-html-components==1.1.3
dash-renderer==1.9.1
dash-table==4.11.3
matplotlib==3.3.2
numpy==1.19.2
numpydoc==1.1.0
pandas==1.1.3

kylegallatin

Kubernetes Practical Intro for Data Scientists

Today we're going to show you how to deploy an application on Kubernetes. You'll only need to Docker Desktop installed locally with Kubernetes enabled.

Testing Our Application with Docker

To deploy an app on Kubernetes, we first need to make sure it works locally with Docker. Use the example fastapi app here: https://github.com/kylegallatin/fast-bad-ml

# clone repo

kylegallatin

import threading
import time
import sys
import asyncio
import concurrent.futures
import time
import nest_asyncio
import pyaudio
import wave