Cancer Cell Classification using Scikit-Learn
Machine Learning is an area of AI that explores how computers can learn without needing to be explicitly taught. As a result, machine learning can help tackle many real-world problems.
<!--more-->
To determine whether cancer cells are malignant or benign, we must examine their physical properties, which we can use to identify them. A machine learning problem will be addressed using Scikit-learn. Scikit-learn, a Python framework, is a free and open-source tool for machine learning, machine mining, and data analysis.
Prerequisites
- Using Scikit-learn will be a necessity for our project.
- Jupyter notebook is preferred for this project. But, you are free to use any IDE of your choice.
- This tutorial will use the Breast Cancer Wisconsin (diagnostic) dataset. Therefore, ensure it's loaded. You can download a copy of this dataset from here.
Table of contents
- Prerequisites
- Table of contents
- Naive Bayes algorithm
- Advantages of the Naive Bayes classification
- Disadvantages of the Naive Bayes classification
- Installations
- The dataset
- Classification implemented with the Scikit-learn framework
- Conclusion
Naive Bayes algorithm
Despite its simplicity, Naive Bayes is a robust algorithm for predicting outcomes. In Naive Bayes, each input variable is assumed to be independent, which is why it's named naive. In reality, this is a big assumption, and the technique works well for a wide range of challenging tasks. Furthermore, when the assumption of independence holds, a Naive Bayes classifier performs better than other models like logistic regression, and you need less training data.
It performs well when you need to use categorical input variables compared to numerical ones. Therefore, this popular method for binary classification, known as Naive Bayes, will be selected for use in this tutorial.
Advantages of the Naive Bayes classification
- It's easy to get started using it.
- Less training data is needed for training.
- Continuous and discrete data are also supported.
- The number of predictors and data points can be increased indefinitely, making it extremely flexible and adaptable.
- It is rapid and can be used to make predictions in real-time.
Disadvantages of the Naive Bayes classification
- In Naive Bayes, all predictors (or traits) are assumed to be independent. However, this is rarely the case in reality. This technique uses the zero-frequency problem to avoid assigning zero probability to a categorical variable whose category was not accessible in the training dataset. Zero frequency indicates simply a continuous term, no wave, no peaks passing.
Installations
Please run the following line on the command prompt to download and install scikit-learn on your computer. You can as well follow this documentation to install it.
pip install scikit-learn
For this project, Jupyter notebook is encouraged, although you can use any IDE you choose. Instead of developing an entire script from scratch, one can execute a few lines of code and see what happens one by one rather than writing the whole script and running it.
To install jupyter notebook, use the command line and enter the following code:
pip install jupyter
The dataset
We will be classifying cancer cells based on their features and identifying if they are malignant or benign using the scikit-learn library available for Python programming language. We will be using the Breast Cancer Wisconsin dataset for our machine learning problem.
The data set and their respective classification labels include malignant and benign breast cancer tumours. The data set has 569 tumors and includes data on 30 attributes of a tumor we will use to train our model. We can use the load_breast_cancer() function to load it.
Classification implemented with the Scikit-learn framework
1. Adding the required modules and data to the import
Scikit-learn and Breast Cancer Wisconsin (diagnostic) dataset will be imported into our program as a first step.
import sklearn
from sklearn.datasets import load_breast_cancer
2. Creating a variable from the dataset
We must consider the labels, classification labels, feature meanings, and the data to be learned from this dataset.
data = load_breast_cancer()
We've loaded the breast cancer dataset into the data variable.
3. Arranging and analyzing data
Before we can learn how to train our model, we need to organize the data. Then, we can use the print() function to see what it contains.
# Organizing our data
namesForlabels = data['target_names']
ourlabels = data['target']
namesForfeature = data['feature_names']
ourfeatures = data['data']
You can then use the print() function to examine the data.
print(namesForlabels)
Output:
['malignant' 'benign']
Tumor data is classified as malignant or non-malignant, as you can see.
Labels 0 and 1 are binary values representing labels where 0 indicates malignant tumors and 1 indicates benign tumors when written.
print(ourlabels)
Output:
[0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
1 0 0 0 0 0 0 0 0 1 0 1 1 1 1 1 0 0 1 0 0 1 1 1 1 0 1 0 0 1 1 1 1 0 1 0 0
1 0 1 0 0 1 1 1 0 0 1 0 0 0 1 1 1 0 1 1 0 0 1 1 1 0 0 1 1 1 1 0 1 1 0 1 1
1 1 1 1 1 1 0 0 0 1 0 0 1 1 1 0 0 1 0 1 0 0 1 0 0 1 1 0 1 1 0 1 1 1 1 0 1
1 1 1 1 1 1 1 1 0 1 1 1 1 0 0 1 0 1 1 0 0 1 1 0 0 1 1 1 1 0 1 1 0 0 0 1 0
1 0 1 1 1 0 1 1 0 0 1 0 0 0 0 1 0 0 0 1 0 1 0 1 1 0 1 0 0 0 0 1 1 0 0 1 1
1 0 1 1 1 1 1 0 0 1 1 0 1 1 0 0 1 0 1 1 1 1 0 1 1 1 1 1 0 1 0 0 0 0 0 0 0
0 0 0 0 0 0 0 1 1 1 1 1 1 0 1 0 1 1 0 1 1 0 1 0 0 1 1 1 1 1 1 1 1 1 1 1 1
1 0 1 1 0 1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1 1 0 1 0 1 1 1 1 0 0 0 1 1
1 1 0 1 0 1 0 1 1 1 0 1 1 1 1 1 1 1 0 0 0 1 1 1 1 1 1 1 1 1 1 1 0 0 1 0 0
0 1 0 0 1 1 1 1 1 0 1 1 1 1 1 0 1 1 1 0 1 1 0 0 1 1 1 1 1 1 0 1 1 1 1 1 1
1 0 1 1 1 1 1 0 1 1 0 1 1 1 1 1 1 1 1 1 1 1 1 0 1 0 0 1 0 1 1 1 1 1 0 1 1
0 1 0 1 1 0 1 0 1 1 1 1 1 1 1 1 0 0 1 1 1 1 1 1 0 1 1 1 1 1 1 1 1 1 1 0 1
1 1 1 1 1 1 0 1 0 1 1 0 1 1 1 1 1 0 0 1 0 1 0 1 1 1 1 1 0 1 1 0 1 0 1 0 0
1 1 1 0 1 1 1 1 1 1 1 1 1 1 1 0 1 0 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1
1 1 1 1 1 1 1 0 0 0 0 0 0 1]
When the features are printed, the tumor's 30 distinct characteristics and properties are seen in full detail. In addition, training our model on factors such as a tumor's malignant or benign status helps us make accurate predictions.
print(ourfeatures)
Output:
[[1.799e+01 1.038e+01 1.228e+02 ... 2.654e-01 4.601e-01 1.189e-01]
[2.057e+01 1.777e+01 1.329e+02 ... 1.860e-01 2.750e-01 8.902e-02]
[1.969e+01 2.125e+01 1.300e+02 ... 2.430e-01 3.613e-01 8.758e-02]
...
[1.660e+01 2.808e+01 1.083e+02 ... 1.418e-01 2.218e-01 7.820e-02]
[2.060e+01 2.933e+01 1.401e+02 ... 2.650e-01 4.087e-01 1.240e-01]
[7.760e+00 2.454e+01 4.792e+01 ... 0.000e+00 2.871e-01 7.039e-02]]
There are 569 instances of tumor data in the dataset in question, each with a numerical value for one of the 30 variables in question. So, from the above data, we can conclude that the first instance of tumor is malignant and it has a mean radius of value 1.79900000e+01.
4. Creating sets of data to organize the information
We need to evaluate our classifier's accuracy by running it on data never seen before. We'll split our data into two sets, the training and test sets, before creating our model. The model's training and evaluation will be done on the training set, and then we will use it to predict the unknown test set.
Sklearn comes with a function called train_test_split(), which automatically splits the data into these two groups for us.
from sklearn.model_selection import train_test_split
train, test, trainingLabels, testingLabels = train_test_split(ourfeatures, ourlabels, test_size = 0.33, random_state = 42)
It is possible to utilize the test size as a parameter to divide the data randomly. We've split the original data into test data by a third of its original size. Other than that, the rest of the data is used for training. A different label for the train and test variables is used for each.
Refer to the documentation for further information on how to use train_test_split() function.
5. Building the Model
You can select from a wide variety of machine learning models to meet your needs. Each one has advantages and cons of its own. A popular method for binary classification, Naive Bayes, has been selected for this model. GaussianNB is the first module that needs to be imported and initialized using the GaussianNB() function and then use the fit() method to train your model on the dataset's data.
from sklearn.naive_bayes import GaussianNB
gaussiannb = GaussianNB()
ourModel = gaussiannb.fit(train, trainingLabels)
To generate predictions on our test set, we first need to train the model. The built-in predict() function returns a prediction value array for each data point in the test set. Next, our forecasts will be displayed using the print() method.
ourPredictions = gaussiannb.predict(test)
print(ourPredictions)
Output:
[1 0 0 1 1 0 0 0 1 1 1 0 1 0 1 0 1 1 1 0 1 1 0 1 1 1 1 1 1 0 1 1 1 1 1 1 0
1 0 1 1 0 1 1 1 1 1 1 1 1 0 0 1 1 1 1 1 0 0 1 1 0 0 1 1 1 0 0 1 1 0 0 1 0
1 1 1 1 1 1 0 1 1 0 0 0 0 0 1 1 1 1 1 1 1 1 0 0 1 0 0 1 0 0 1 1 1 0 1 1 0
1 1 0 0 0 1 1 1 0 0 1 1 0 1 0 0 1 1 0 0 0 1 1 1 0 1 1 0 0 1 0 1 1 0 1 0 0
1 1 1 1 1 1 1 0 0 1 1 1 1 1 1 1 1 1 1 1 1 0 0 1 1 0 1 1 0 1 1 1 1 1 1 0 0
0 1 1]
6. Accuracy testing of the model trained
If we want to make sure our model is accurate, we may compare its predictions to the actual labels in the testing set. We use the sklearn module built-in accuracy_score() function for this task.
from sklearn.metrics import accuracy_score
print(accuracy_score(testingLabels, ourPredictions))
Output:
0.9414893617021277
These results show that this machine learning classifier based on the Naive Bayes algorithm is 94.15% accurate.
To see the complete code for this tutorial, click here.
Conclusion
Determination can be done by looking at the features of cancer cells, which can indicate whether cancer cells are malignant or benign. We used scikit-learn to address this machine learning problem. Python framework Scikit-learn is a free and open-source tool for machine learning, data mining, and data analysis.
Peer Review Contributions by: Willies Ogola
