Data Science

Case Study of Amazon Review System

So, First of all we need to understand the ‘ Traditional programming ’ and ‘Machine Learning Programming’.

What is Machine Learning Programming and Traditional Programming ?

Lets talk about first :-

1.Traditional Programming:-

In simple way I just want to say , In Traditional Programming all work is doing in manual form.
In traditional work, Men work will be more, It will be costly as well as lengthy.
Due to increasing of data day-by-day ,this programming work will not work.
That’s why, Machine Learning programming has started to be used.

2. Machine Learning Programming:-

Machine learning programming is nothing but it is a prediction machine, In which you feed input data to train the algorithm and that algorithm make a model.
In that model, We feed our output to know the accuracy of our model.

“ In short, Machine Learning is a combination of human brain, ideas and It’s own strategy to predict the data.”

How to Predict the data by using Machine learning algorithm.?

Lets take a example :-
“Amazon Review System”
As we already know Traditional Programming and Machine learning programming.
First of all , We have to know the Problem Statement of this case study.
We want the best product in-front of the page in Amazon platform.
There are so many products in Amazon , But we have to take a example of ‘T-shirts’ only. In daily life ,we can see that there are lots of people buy those products and gives their reviews.

Why is Amazon review analysis important.?

Fig:- (2) Amazon Review Analysis

PROCEDURE ON HOW MACHINE LEARNING MODEL WORK :

Let’s take a sample data From Amazon , We just split the data into training data and testing data.

Fig:-(3) Amazon Sample Data

Why we need splitting of data.?

Because, You can’t evaluate the predictive performance of a model by using same data that you’ve use for training.
To know the accuracy of model we need some fresh data.
That’s why we splitting data.
So, we splitting the data into 80% of ‘Training Data’ and 20% of ‘Testing Data’.

Fig:-(4) Splitting of Data Set

First of all talk about Training Data, Training data is nothing but the sample of that data we can use to train the machine learning algorithm to make a model.
Lets take a example of cooking, In cooking — The ingredients and the procedure of cooking is our training data.
After the working on training data or we can say working on our cooking , Before serving the dish to guest we will taste that dish as a ‘Actual Testing Data’ and then serve the guest. At the end compare your opinion about your dish with guest opinions and then you know the accuracy of your dish.
Like that, We take X-Train Data as a Input and Y-Train Data as a Output. In that Input is called as ‘Features’, ’Content’ or ’Review’ and Output is called as ‘Class’ or ‘Label’.
So, After that we feed the Training input data as a ‘Review’ and output data as a ‘Decision’ according to fig:-(1). To train the machine learning algorithm to make a Model or we can say Function .

Fig:-(5) Working on Training Data

After that process, We feed the testing input data (X-test)in model, Our actual test output data(Y-test) not be available to my function.

Fig:-(6) Testing Model (When Function Is Ready)

After that whole process we compared the Actual Data(Y-actual data) with Predicted data(Y-Predicted data) To know the accuracy of our Model

Fig:-(7) Comparison Chart

In that we found only two matching output , it means the accuracy of our model is 40%.
In this way we can see the accuracy of our model.

Why we doing this whole process.?

To make sure that our model is working perfectly in the real data of real world.
Machine Learning never give 100% result, In that we only want a prediction
So , In this way we know the accuracy and find out the prediction of our sample data to make sure that our model is working exactly in real world data.

Thank You!…..

Please login in order to make a comment.