Machine learning is basically a subset of artificial intelligence that uses previously existing data to make a prediction on new data. Of course, all of us know this by now! This article demonstrates how a machine learning model developed in Python can be used as a part of a Java code to make predictions.
This article assumes you are familiar with the basic development skills and understanding of machine learning. We will start with training our model, and then make a machine learning model in Python.
This article assumes you are familiar with the basic development skills and understanding of machine learning. We will start with training our model, and then make a machine learning model in Python.
I am taking the example of a flood prediction model. First, import the following libraries:
import pandas as pd import numpy as np import matplotlib.pyplot as plt
Once we have successfully imported the libraries, we need to take in the data sets, as shown in the code below. To predict floods, I am using the river level data set.
from google.colab import files
uploaded = files.upload()
for fn in uploaded.keys(): print(‘User uploaded file “{name}” with length {length} bytes’.format(
name=fn, length=len(uploaded[fn])))
Choose files No file chosen
The upload widget is only available when the cell has been executed in the current browser session. Please rerun this cell to enable Saving Hoppers Crossing-Hourly-River-Level.csv to Hoppers Crossing-Hourly-River-Level.csv User uploaded file “Hoppers Crossing-Hourly-River-Level.csv” with length 2207036 bytes.
Once this is done, we can train our model using the sklearn library. For this, we first need to import the library and the algorithm model, as shown in Figure 1.
from sklearn.linear_model import LinearRegression regressor = LinearRegression() regressor.fit(X_train, y_train)
Once that is done we have trained our model, and it’s now ready to make predictions, as shown in Figure 2.
Using ML model in Java
What we need to do now is to convert the ML model into a model that can be used by a Java program. There is a library called sklearn2pmml that helps us do this:
Install the library, pip install sklearn2pmml
Once the library is installed we can convert our already trained model, as shown below:
sklearn2pmml(pipeline, ‘model.pmml’, with_repr = True)
This is it! We can now use the generated model.pmml file in our Java code to make predictions. Do try it out!
