Deploy a Machine Learning Model using Streamlit Library

Machine Learning:

Machine learning is the ability of a computer to learn from its experience, even if it has not been programmed. Machine Learning is a hot field right now, and many top companies around the globe are using it to improve their products and services. A Machine Learning model that is not trained in our Jupyter Notebook is useless. We need to make these models available for everyone so they can be used.

In this tutorial we will train an Iris Species classification classifier and then deploy the model with Streamlit, an open-source app framework that allows us to deploy ML models easily.

Streamlit Library:

Streamlit allows us to create apps for our machine-learning project with simple Python scripts. Hot reloading is also supported, so our app can be updated live while we edit and save our file. Streamlit API allows us to create an app in a few lines of code (as we’ll see below). Declaring a variable is the same thing as adding a widget. We don’t need to create a backend, handle HTTP requests or define different routes. It’s easy to set up and maintain.

First, we will train the model. As the primary purpose of this tutorial, we will not be doing much pre-processing.

Required Modules and Libraries:

First, we must install the following:

  !pip3 install pandas  !pip3 install numpy  !pip3 install sklearn  !pip3 install streamlit  

Dataset:

  import pandas as pnd  import numpy as nmp      dataframe1 = pnd.read_csv(‘iris.csv’)  dataframe1.head()    

Output:

Deploy a Machine Learning Model using Streamlit Library

We will now drop the Id column as it is not necessary for Iris species classification. Next, we will divide the data into a training and testing set and use a Random Forest Classifier. Any other classifier can also be used, such as logistic regression or support vector machine.

Code:

  # Now, we will be dropping the Id column  dataframe1.drop(‘Id’, axis = 1, inplace = True)  dataframe1.head()  # Here, we will be renaming the target column into the numbers for aid training of the model  dataframe1[‘Species’] = dataframe1[‘Species’].map({‘Iris-setosa’:1, ‘Iris-versicolor’:2, ‘Iris-virginica’:3})    # Now, we will split the data into the columns which would need to be trained (A) and the target column (b)  A = dataframe1.iloc[:, :-1]  b = dataframe1.iloc[:, -1]    # After splitting the data into training and testing data with 30 % of data as test data respectively  from sklearn.model_selection import train_test_split as tts  A_train, A_test, b_train, b_test = tts(X, y, test_size = 0.3, random_state = 0)    # Now, we will import the random forest classifier model and train it on the dataset  from sklearn.ensemble import RandomForestClassifier as RFC  classifier1 = RFC()  classifier1.fit(A_train, b_train)    # Here, we will predict on the testing dataset  b_pred = classifier1.predict(A_test)    # At last we will find out the accuracy of our prediction  from sklearn.metrics import accuracy_score as a_s  score = a_s(b_test, b_pred)  print (“Our Prediction Accuracy is: “, score)  

Output:

Our Prediction Accuracy is:  0.9777777777777777

As, we got the accuracy of 97.77%, which is quite good.

To use this model for predicting unknown data, we must save it. A pickle is a tool that serializes and deserializes a Python object structure.

Code:

  alizes a Python object structure.  Code:  # Here, we will pickle the model  import pickle as pkl  pickle_out1 = open(“classifier1.pkl”, “wb”)  pkl.dump(classifier1, pickle_out1)  pickle_out1.close()  

A new file called “classifier1.pkl”, will be created in the same directory. We can now use Streamlit to deploy our model –

Copy the code below into another Python file.

Code:

  import pandas as pnd  import numpy as nmp  import pickle as pkl  import streamlit as smt  from PIL import Image as img      # loading in the model to predict on the data  pickle_in1 = open(‘classifier1.pkl’, ‘rb’)  classifier1 = pkl.load(pickle_in1)      def welcome():      return ‘welcome you all’      # here, we will define the function which will make the prediction using the      # data which the user have imported  def prediction1(sepal_length1, sepal_width1, petal_length1, petal_width1):             prediction1 = classifier1.predict(          [[sepal_length1, sepal_width1, petal_length1, petal_width1]])      print(prediction1)      return prediction1              # Here, this is the main function in which we will be defining our webpage   def main():        # Now, we will give the title to out web page      smt.title(“Iris Flower Prediction”)              # Now, we will be defining some of the frontend elements of our web            # page like the colour of background and fonts and font size, the padding and    # the text to be displayed      html_temp = “””      <div style = “background-colour: #FFFF00; padding: 16px”>      <h1 style = “color: #000000; text-align: centre; “> Streamlit Iris Flower Classifier ML App        </h1>      </div>      “””              # Now, this line will allow us to display the front-end aspects we have       # defined in the earlier      smt.markdown(html_temp, unsafe_allow_html = True)              # Here, the following lines will create the text boxes in which the user can     # enter the data which is required for making the prediction      sepal_length1 = smt.text_input (“Sepal Length “, ” Type Here”)      sepal_width1 = smt.text_input (“Sepal Width “, ” Type Here”)      petal_length1 = smt.text_input (“Petal Length “, ” Type Here”)      petal_width1 = smt.text_input (“Petal Width “, ” Type Here”)      result = ” ”              # here, the below line will ensure that whenever the button named ‘Predict’ # is clicked, the prediction function that is defined earlier is called for making   # the prediction and it will also store it in the variable result      if smt.button (“Predict”):          result = prediction1 (sepal_length1, sepal_width1, petal_length1, petal_width1)      smt.success (‘The output of the above is {}’.format(result))  if __name__== ‘__main__’:      main()  

This command can be executed by entering the following command into the terminal:

Output:

Deploy a Machine Learning Model using Streamlit Library

app1.py is where the Streamlit code was written.

After the website opens in our browser, we can then test it. We can also use this method to deploy deep learning and machine-learning models.

Next TopicDifferent Types of Methods for Clustering Algorithms in ML

Deploy a Machine Learning Model using Streamlit Library

Deploy a Machine Learning Model using Streamlit Library

Machine Learning:

Streamlit Library:

Required Modules and Libraries:

Dataset:

Anti-Money Laundering using Machine Learning

Basic Configuration in Magento 2

You may also like