Steps to perform Machine Learning
Welcome Fellas! Let's start with machine learning steps. Training a model for Machine Learning needs data. We will see the process which we follow to train a machine learning model, below are the steps required for training a model:-
1. Identify Problem:- The first step in any project is to identify a problem. All your effort will be zero if you are solving a wrong problem. So the first task is to identify the problem. Once you have your problem ready you are ready to go to next step.
2. Data Analysis:- For solving any problem you first need to analyze what you have to solve the problem, so in case of machine learning you have data with you as a problem set. You need to understand this data so that you have better understanding of data. To understand data better you can either visualize your data by creating charts or you can summarize data by checking the different aspects of data like mean, median, missing values, wrong values etc.
As you are starting to learn machine learning you need data. Data can be collected from public domains, government sites etc. You can even create your own dataset by using web scraping. Data can be csv file, excel, images, videos, mp3, text file, pdf etc.
You can download data from Kaggle, this link for public data or github collection of public datasets.
3. Data preprocessing:- Machine Learning Algorithms learn from data. It is necessary that you feed right data to algorithm to solve the problem right way. You need to check what data is available to you, what features are present in data, is data properly formatted. By the end of this step you data will be ready to server for machine learning algorithm.
Most of the data we have in real world is not complete. Most of the have missing values, wrong values, inconsistent values, extreme values in data. In this step we process the data to make it clean and convert it to proper format so that we can feed this for training model. Usually this steps takes most of the time while training a model as cleaning data will take most of the time.
4. Training Model:- This is the step where actual machine learning happens. You have processed data, you feed it to algorithm to train a model, you test your model, make tweaks, try different algorithms. There are lot of machine learning algorithms and you need to choose which machine learning algorithm suits your data best. You can try a bunch of different algorithms and then compare the results to see which algorithm is best for your data.
5. Cross Validation/Testing:- In this step you test your trained model. You feed model with the data which model has not seen yet and then analyze the output to check the errors. We use different data for testing then the data we use for training. With this step you also analyze if your model is over-fitting or under-fitting your data.
6. Present Result/Visualization:- In this step we see the describe our problem and solution in such a way so that business people/third party can understand the results. You can create case studies, graphs, user stories to present the problem and solution/observations for better understanding.
So we have seen the steps we need to perform in order to train create a machine learning model. Next time we will be seeing how you can setup your machine learning environment and then we will create a simple machine learning program to get started.
Till then Enjoy Machine Learning.
1. Identify Problem:- The first step in any project is to identify a problem. All your effort will be zero if you are solving a wrong problem. So the first task is to identify the problem. Once you have your problem ready you are ready to go to next step.
2. Data Analysis:- For solving any problem you first need to analyze what you have to solve the problem, so in case of machine learning you have data with you as a problem set. You need to understand this data so that you have better understanding of data. To understand data better you can either visualize your data by creating charts or you can summarize data by checking the different aspects of data like mean, median, missing values, wrong values etc.
As you are starting to learn machine learning you need data. Data can be collected from public domains, government sites etc. You can even create your own dataset by using web scraping. Data can be csv file, excel, images, videos, mp3, text file, pdf etc.
You can download data from Kaggle, this link for public data or github collection of public datasets.
3. Data preprocessing:- Machine Learning Algorithms learn from data. It is necessary that you feed right data to algorithm to solve the problem right way. You need to check what data is available to you, what features are present in data, is data properly formatted. By the end of this step you data will be ready to server for machine learning algorithm.
Most of the data we have in real world is not complete. Most of the have missing values, wrong values, inconsistent values, extreme values in data. In this step we process the data to make it clean and convert it to proper format so that we can feed this for training model. Usually this steps takes most of the time while training a model as cleaning data will take most of the time.
4. Training Model:- This is the step where actual machine learning happens. You have processed data, you feed it to algorithm to train a model, you test your model, make tweaks, try different algorithms. There are lot of machine learning algorithms and you need to choose which machine learning algorithm suits your data best. You can try a bunch of different algorithms and then compare the results to see which algorithm is best for your data.
5. Cross Validation/Testing:- In this step you test your trained model. You feed model with the data which model has not seen yet and then analyze the output to check the errors. We use different data for testing then the data we use for training. With this step you also analyze if your model is over-fitting or under-fitting your data.
6. Present Result/Visualization:- In this step we see the describe our problem and solution in such a way so that business people/third party can understand the results. You can create case studies, graphs, user stories to present the problem and solution/observations for better understanding.
So we have seen the steps we need to perform in order to train create a machine learning model. Next time we will be seeing how you can setup your machine learning environment and then we will create a simple machine learning program to get started.
Till then Enjoy Machine Learning.
No comments