CardamomAI Demo Page

import argparse
import numpy as np
import pandas as pd
from json import dumps
import sys
import time

import requests
from sklearn.datasets import load_iris
from sklearn.model_selection import train_test_split
from sklearn.neural_network import MLPClassifier
from sklearn.tree import DecisionTreeClassifier
from sklearn.naive_bayes import GaussianNB

from cardamom_ai import deploy_model

def model_to_classifier(model):
    if model == "decision_tree" or model == "dt":
        return DecisionTreeClassifier(max_depth=3, random_state=42)
    elif model == "neural_network" or model == "nn":
        return MLPClassifier(random_state=42)
    elif model == "gaussian_naive_bayes" or model == "nb":
        return GaussianNB()
    else:
        return None

def pretty_model_output(classes, data):
    return [dict(sorted(zip(classes, data), key=lambda x: x[1], reverse=True))]

if __name__ == "__main__":
    parser = argparse.ArgumentParser(description='Pickles an example ML Model')
    parser.add_argument('model', type=str, help='Which model type to generate')
    parser.add_argument('name', type=str, help="Name of model to deploy")
    args = parser.parse_args()
    clf = model_to_classifier(args.model)
    if clf is None:
        print("Model not supported!", file=sys.stderr)
        sys.exit(1)

    data = load_iris()
    df = pd.DataFrame(data.data, columns=data.feature_names)
    
    df['Species'] = data.target
    target, target_names = np.unique(data.target), np.unique(data.target_names)
    df['Species'] = df['Species'].replace(dict(zip(target, target_names)))
    
    x = df.drop(columns='Species')
    y = df['Species']
    X_train, x_test, y_train, y_test = train_test_split(x,
                                                        y,
                                                        test_size = 0.4,
                                                        random_state = 42)
    print("Training model")
    clf.fit(X_train, y_train)
    print("Deploying model")
    predict_endpoint = deploy_model(clf, args.name)
    if not isinstance(predict_endpoint, str):
        sys.exit(predict_endpoint)
    time.sleep(10)
    print("New API available at: ", predict_endpoint)
    for _ in range(5):
        time.sleep(2)
        ind = np.random.randint(0, len(x_test))
        print("Python: \n", pretty_model_output(
            clf.classes_.tolist(),
            clf.predict_proba(x_test.values[ind].reshape(1, -1))[0].tolist()
        ))
        res = requests.post(predict_endpoint, json=x_test.iloc[ind].to_dict())
        print("API: \n", dumps(res.json(), separators=(': ', ', ')))

Hi HN!

Save the script to the left under a file name, I use demo.py
pip install cardamom-ai pandas
export CARDAMOM_HOST="api.cardamom.ai"
Run the saved file, for example python3 demo.py decision_tree test-model
Note the second argument is the model algorithm, the third is the deployed name
Enjoy deploying your model to AWS Lambda

Since this is just a demo I'll probably delete the data after this weekend, so if you want hosting on a sustained basis please don't hesitate to sign up for the email list or reach out via my profile. :)

Below I've included a video for phone users to check it out