How can I run an ML algo in a web browser using an API interface or Flask?

Question

I am just starting to get into Flask. I created a super-simple Flask app.

from flask import Flask

app = Flask(__name__)

@app.route("/")
def home():
    return "Hello, World!"
    
if __name__ == "__main__":
    app.run(debug=True)

I can run 'python hello-world.py' in my Anaconda prompt, go to 'localhost:5000' and it works as expected. I'm wondering how I can use some ML code to create some kind of API interface, using Flask. This is just a learning exercise for me, so I don't really have a concrete example of what needs to be done. I'm calling this a 'deployment exercise'. Anyway, here is a sample of code that I want to run.

# K-Means
from sklearn.cluster import KMeans
import matplotlib.pyplot as plt
from mpl_toolkits.mplot3d import Axes3D
import numpy as np
# %matplotlib inline
from sklearn import datasets#Iris Dataset
iris = datasets.load_iris()
X = iris.data#KMeans
km = KMeans(n_clusters=3)
km.fit(X)
km.predict(X)
labels = km.labels_#Plotting
fig = plt.figure(1, figsize=(7,7))
ax = Axes3D(fig, rect=[0, 0, 0.95, 1], elev=48, azim=134)
ax.scatter(X[:, 3], X[:, 0], X[:, 2],
          c=labels.astype(np.float), edgecolor="k", s=50)
ax.set_xlabel("Petal width")
ax.set_ylabel("Sepal length")
ax.set_zlabel("Petal length")
plt.title("K Means", fontsize=14)

# GMM:
from sklearn.mixture import GaussianMixture
import matplotlib.pyplot as plt
from mpl_toolkits.mplot3d import Axes3D
import numpy as np
#%matplotlib inline
from sklearn import datasets#Iris Dataset
iris = datasets.load_iris()
X = iris.data#Gaussian Mixture Model
gmm = GaussianMixture(n_components=3)
gmm.fit(X)
proba_lists = gmm.predict_proba(X)#Plotting
colored_arrays = np.matrix(proba_lists)
colored_tuples = [tuple(i.tolist()[0]) for i in colored_arrays]
fig = plt.figure(1, figsize=(7,7))
ax = Axes3D(fig, rect=[0, 0, 0.95, 1], elev=48, azim=134)
ax.scatter(X[:, 3], X[:, 0], X[:, 2],
          c=colored_tuples, edgecolor="k", s=50)
ax.set_xlabel("Petal width")
ax.set_ylabel("Sepal length")
ax.set_zlabel("Petal length")
plt.title("Gaussian Mixture Model", fontsize=14)

If my approach is totally wrong, please tell me how to do this the right way. Again, this is simply a learning exercise for myself. I've been using Spyder to do all my ML work for a while, and now I want to try to incorporate some API technology, or Flask technology, into my work.

Answer 1

You can save models in most of the libraries, including sckit-learn. For example

import pickle
pickle.dump(kmeans, open("kmean_model.pkl", "wb"))

and load it as

kmeans = pickle.load(open("kmean_model.pkl", "rb"))

You can use Api endpoint to use this loaded model and predict, for example

from flask import Flask, jsonify

import pickle
import pandas as pd
app = Flask(__name__)
@app.route('/predict', methods=['POST'])
def predict():
     json_features = request.json
     query_df = pd.DataFrame(json_features)
     features = pd.get_dummies(query_df)
    
     prediction = kmeans.predict(features)
     return jsonify({'prediction': list(prediction)})
if __name__ == '__main__':
     kmeans = pickle.load(open("kmean_model.pkl", "rb"))
     app.run(port=8080)

Note: This is overly simplistic example with assumptions, you need to see if you have same features or not and more checks.