Linear regression is a commonly used algorithm used in statistics and machine learning, usually deployed for predictive analysis. It is used to predict the value of one variable (dependent variable), based on the value of another (independent variable).
The popularity of linear regression can be attributed to its simplicity and easy-to-interpret mathematical formulas.
Linear regression can be performed in several programming languages and environments:
- Excel linear regression
- Linear regression Python
- Sklearn linear regression
- MATLAB linear regression
- R linear regression
Today, we list some online courses to learn linear regression from:
Deep Learning Foundation- Linear regression and statistics: Udemy
Through this course, students learn about the Mathematics behind R-Squared, linear regression and VIF, the statistical background of linear regression and assumptions, and assumptions of linear regression hypothesis testing. Additionally, students will be able to develop a deep understanding of Gradient descent and optimisation; they will be able to programme their own version of a linear regression model in Python and learn writing codes for T-test, Z-test and Chi-squared test in Python.
Divided into 44 lectures across eight sections, the course is six and a half hours long. The course fee is Rs 385 on Udemy.
For more information, click here.
Understanding and applying linear regression: Pluralsight
Vitthal Srinivasan designs this beginner-level course, divided into nine chapters:
- Modeling relationships between variables using regression
- Understanding simple regression models
- Implementing simple regression models in Excel
- Implementing simple regression models in R
- Implementing simple regression models in Python
- Understanding multiple regression models
- Implementing multiple regression models in Excel
- Implementing multiple regression models in R
- Implementing multiple regression models in Python
It approximately takes four hours to complete this course.
For more information, click here.
Linear regression and modeling: Coursera
Offered by Duke University, Mine Cetinkaya-Rundel delivers this beginner-level course. Learners are introduced to the concepts of simple and multiple linear regression models. It is divided into five chapters — introduction to linear regression and modeling; predicting or evaluating the relationship between two numerical variables; outliers, inference in linear regression and variability partitioning; multiple regression; and finally, a project on data analysis.
It takes approximately 10-hours to complete, and on completion, learners receive a shareable certificate.
For more information, click here.
Data science: Linear regression: edX
HarvardX offers this introductory course on using R to implement linear regression. Rafael Irizarry teaches the class.
Upon completing this course, one will learn about the origin of linear regression by Galton, confounding and how to detect it, and examining relationships between variables by implementing linear regression in R.
Learners can audit the course for free. However, for a verified certificate, one needs to pay $99 or Rs 7,336. The self-paced course approximately takes eight weeks to complete, provided one commits one to two hours every week.
For more information, click here.
Correlation and regression in R: Datacamp
By completing this course, students will learn to describe relationships between two numerical quantities and characterise the relationship between them graphically — in the form of summary statistics and through linear regression models.
The course is divided into five modules:
- Visualising two variables: learning techniques for exploring bivariate relationships
- Correlation as a means of quantifying bivariate relationships
- Simple linear regression
- Interpreting regression models
- Assessing the fit of linear regression models
The course is designed and delivered by Ben Baumer, Assistant Professor at Smith College.
The prerequisite for taking this course includes having knowledge of or completing a course on exploratory data analysis in R.
For more information, click here.
Deep learning prerequisites – Linear regression in Python: Udemy
This course teaches how to derive and solve linear regression models and apply it appropriately to data science problems. Additionally, they will be able to programme their own version of the linear regression model in Python. Students will get to learn about 1-D linear regression, multiple linear regression and polynomial regression, practical machine learning issues, and setting up one’s environment.
Consisting of 53 lectures, this course is spread across nine sections and is approximately six hours long. For this course, the prerequisites include knowledge of basic Python programming and derivation using calculus.
For more information, click here.
Linear regression is mainly used to forecast trends and effects and to quantify cause-effect relationships. Linear regression’s real-time applications include analysing pricing elasticity, predicting trends and sales estimates, for sports analysis and assessing insurance risks. Thus, it can be used as a scientific and reliable method to predict the future and train models quickly.