MLops : Session - 15
In the last article, I explain how we do feature selection and feature elimination using Visuals(Graphs). We also see some data cleaning part which is almost 70-80% work of a data scientist. Here I mention one more thing that for better doing feature selection task, you must know a little bit of that domain.
Today I start model training and prediction part and also know how to check accuracy in binary classification.
Binary Classification:
Let me start where I left in the last session.
After doing the data cleaning part, we get this clean dataset which is ready for training. But we also have some categorical variables in our dataset, so we also do ONEHOT encoding on our categorical variable. I discuss this in this article.
Now I got my independent and dependent variable. Now come to the next step.
Training and Prediction part:
We use Logistic Regression to create and train our model.
Now our model is ready for prediction.
Now come to the most important part.
Check the accuracy of the model:
For checking the accuracy of our model, we use Confusion_matrix.
By the above figure, you easily understand what happens. We use this matrix to get the model accuracy.
record = 140+17+33+78 trueanswer = 140 + 78 accurcy = trueanswer / record * 100 error = 33 + 17 error% = error / record *100
Finally, you get that your model is 80% accurate.😊
In our next article, we see some more things...
Hope you learn something. I suggest do one hands-on practice too for better understanding,
Happy Learning :)
