Machine learning is the application of Artificial Intelligence where a predictive model learns identifying pattern from the data. I have prepared a set of articles with practical demonstration on how you can learn Machine Learning and build your own predictive models. All of these articles are published in Analytics India Magazine and Analytics Steps website.
Analytics India Magazine Profile - https://analyticsindiamag.com/author/rohit-dwivedianalyticsindiamag-com/
Analytics Steps Profile - https://www.analyticssteps.com/author/rohit-dwivedi
Connect with me on Linkedin - https://www.linkedin.com/in/rohit-dwivedi-b5141b148/
• Exploratory Data Analysis
• Machine Learning Algorithms
• Model Performance and Hyperparameter Tuning
• Model Evaluation and Error Metrics
• Regularization
• Machine Learning Model Deployment
• Miscellaneous Articles
• Projects
In the journey of Data science, I feel its the transformation of Data that can help you to attain good results. The data cannot be directly fed to a predictive model and it's very important to pre-process it. Real experience in analytics can come by understanding your data well and the problem that is to be solved.
Do keep this thing in mind- "Garbage In Garbage Out". The below articles will help you to understand more about your data and how to preprocess it in an good manner. The article will help you undetstand more about data analysis.
1) How to do Exploratory Data Analysis before building Machine Learning models?
2) How to do Exploratory Data Analysis Using Pandas Profiling?
3) Outlier Detection Using z-Score – A Complete Guide With Python Codes
Now once we are done with preparing our data now its time to build the predictive model. In machine learning we have several different algorithm to build classification as well as regression models. We need to pick that algorithm that works well for our data. Refer to the below articles that will help you to understand different machine learning algorithms that can be used to build models.
1) How Does K-nearest Neighbor Works In Machine Learning Classification Problem?
2) How Does Linear And Logistic Regression Work In Machine Learning?
3) What Is Naive Bayes Algorithm In Machine Learning?
4) How Does Support Vector Machine Algorithm Works In Machine Learning?
5) Introduction To Principal Component Analysis In Machine Learning
6) Introduction To Decision Tree Algorithm In Machine Learning
7) How to use the Random Forest classifier in Machine learning?
8) What is LightGBM Algorithm, How to use it?
9) Introduction to XGBoost Algorithm for Classification and Regression
10) Random Forest Vs XGBoost – Comparing Tree-Based Algorithms (With Codes)
After we build our model now its time to check the actual power of the model. We can check this using different evulation metrics like accuracy score, confusion matrix, etc. Also, if we want to test how will the model perform on unseen data we make use of techniques like Cross Validation and Boot Strap Sampling. This gives us an idea how will the model perform on production data.
1) What are Model Parameters and Evaluation Metrics used in Machine Learning?
2) Practical Guide to Machine Learning Model Evaluation and Error Metrics
3) Hands-On Implementation of K-Fold Cross-Validation and LOOCV in Machine Learning
4) Hands-On Guide To BootStrap Sampling For ML Performance Evaluation
5) ROC-AUC Curve For Comprehensive Analysis Of Machine Learning Models
6) What is Imblearn Technique – Everything To Know For Class Imbalance Issues In Machine Learning
In machine learning we always define two things one is parameter and other is hyperparamter. The paramter is internal to the model like coefficients in Logistic Regression whereas Hyperparameter are the ones that control the power of an algorithm or an model for example defining k in KNN algorithm is an example of hyperparamter. Check the below articles where you can learn how to tune the model hyperparameter and get the best performance of the machine learning model.
1) Introduction to Model Hyperparameter and Tuning in Machine Learning
2) Guide To Hyperparameters Tuning Using GridSearchCV And RandomizedSearchCV
Regularization is a method that is used to prevent overfitting in machine learning model. This means when the model performs very good in the training but poorly on unseen data. Then we make use of regularization to prevent that. Check the implementation Ridge and Lasso Regression that is widely used regularization technique in machine learning.
1) Regularization in Machine Learning - Hands-On-Implementation of Lasso and Ridge Regression
Once we are done with building models, evaluating the performance and all set to go for unseen data prediction. Now its time to check the real potential of the model by deploying it in real time. Check the below article where I have explained about Deploying a machine learning model using Flask.
1) Hands-On-Guide To Machine Learning Model Deployment Using Flask
2) Complete Tutorial on Tkinter To Deploy Machine Learning Model
1) Step-by-Step Building Block For Machine Learning Models
2) How To Implement ML Models With Small Datasets
3) How Does The Data Size Impact Model Accuracy?
4) Pycharm IDE For Dummies- Beginners Guide
-
In this project we have to build a Classification model that will be able to classify patients as abnormal and normal from the biomechanical features.
-
This Project involved using classification algorithms and ensemble techniques to diagnose Parkinson’s Disease (PD) using the patient voice recording data. Various models were used including Naive Bayes, Logistic Regression, SVM, Decision Tree, Random Forest etc. and comparison of accuracy across these models was done to finalize the model for prediction
-
This project involved prediction of diabetic patients using classification algorithms like Logistic regression, kNN, Naive Bayes and SVM. Exploratory data analysis was done to understand the patterns of data followed with model implementation.
-
Analyzed cars dataset and performed exploratory data analysis and then categorized them using K means clustering. Used linear regression on the different clusters and estimated coefficients.