What do you think?
Rate this book
Linear Regression And Correlation: A Beginner's Guide
Linear Regression & Correlation
If you are looking for a short beginners guide packed with visual examples, this book is for you.
Linear Regression is a way of simplifying a group of data into a single equation. For instance, we all know Moore’s that the number of transistors on a computer chip doubles every two years. This law was derived by using regression analysis to simplify the progress of dozens of computer manufacturers over the course of decades into a single equation. Correlation is a way of calculating how much two sets of numbers change together. In addition to being part of the regression analysis, correlation is heavily used in investment industries, for instance, to determine if two stocks are likely to change value together or independently.
This book goes through how to calculate correlation and linear regression and works through multiple examples of how to do it. Just as importantly, this book is loaded with visual examples of what correlation is and how to use linear regression. This book doesn't assume that you have prior in-depth knowledge of statistics or that you regularly use an advanced statistics software package. If you know what an average is and can use Excel, this book will build the rest of the knowledge, and do so in an intuitive way.
This Is Not A Textbook
The reason I wrote this book is that there are not a lot of good examples out there of how to do multiple regression, which is regression analysis between more than two variables. I checked half a dozen different sources, including several textbooks, on how to do multiple regression. In each case, the source had a lot of information; in some cases dozens of pages; on how to do the data preparation, how to interpret results, and potential problems to watch out for, but to actually do the multiple regression calculation, they all said to use a software package, like Matlab or Minitab.
To put it a different way, those sources are a lot like a driver’s education class teaching you about setting the mirrors, adjusting the seat, and fastening the safety belt - but just when you think you are finally ready to actually start the car and put it in gear, they tell you to take the bus.
This book is different. It does not spend much time on the niceties of exactly how you should scrub your data, and instead just shows how to do the calculations, with examples. This book is more like Grandpa showing you how to actually drive the old pickup truck from the farmhouse to the barn so you can get some work done.
What Is In This Book?
There are a number of examples shown in this book, they include
How to do a correlation calculationAn example of correlation on the stock price of 10 different big-name stocks, such as Coke and PepsiHow having uncorrelated investments can give you better returns at lower risk.How to do linear regression with two variablesHow to do multiple linear regression with any number of independent variablesA regression analysis to predict the number of viewers in future episodes of the television show ‘Modern Family’How to evaluate the quality of your regression analysis using R-squared or adjusted R-squaredHow to do regression on exponential data, and recreate Moore’s law
If you are a visual learner and like to learn by example, this intuitive book might be a good fit
If you are looking for a short beginners guide packed with visual examples, this book is for you.
Linear Regression is a way of simplifying a group of data into a single equation. For instance, we all know Moore’s that the number of transistors on a computer chip doubles every two years. This law was derived by using regression analysis to simplify the progress of dozens of computer manufacturers over the course of decades into a single equation. Correlation is a way of calculating how much two sets of numbers change together. In addition to being part of the regression analysis, correlation is heavily used in investment industries, for instance, to determine if two stocks are likely to change value together or independently.
This book goes through how to calculate correlation and linear regression and works through multiple examples of how to do it. Just as importantly, this book is loaded with visual examples of what correlation is and how to use linear regression. This book doesn't assume that you have prior in-depth knowledge of statistics or that you regularly use an advanced statistics software package. If you know what an average is and can use Excel, this book will build the rest of the knowledge, and do so in an intuitive way.
This Is Not A Textbook
The reason I wrote this book is that there are not a lot of good examples out there of how to do multiple regression, which is regression analysis between more than two variables. I checked half a dozen different sources, including several textbooks, on how to do multiple regression. In each case, the source had a lot of information; in some cases dozens of pages; on how to do the data preparation, how to interpret results, and potential problems to watch out for, but to actually do the multiple regression calculation, they all said to use a software package, like Matlab or Minitab.
To put it a different way, those sources are a lot like a driver’s education class teaching you about setting the mirrors, adjusting the seat, and fastening the safety belt - but just when you think you are finally ready to actually start the car and put it in gear, they tell you to take the bus.
This book is different. It does not spend much time on the niceties of exactly how you should scrub your data, and instead just shows how to do the calculations, with examples. This book is more like Grandpa showing you how to actually drive the old pickup truck from the farmhouse to the barn so you can get some work done.
What Is In This Book?
There are a number of examples shown in this book, they include
How to do a correlation calculationAn example of correlation on the stock price of 10 different big-name stocks, such as Coke and PepsiHow having uncorrelated investments can give you better returns at lower risk.How to do linear regression with two variablesHow to do multiple linear regression with any number of independent variablesA regression analysis to predict the number of viewers in future episodes of the television show ‘Modern Family’How to evaluate the quality of your regression analysis using R-squared or adjusted R-squaredHow to do regression on exponential data, and recreate Moore’s law
If you are a visual learner and like to learn by example, this intuitive book might be a good fit
- GenresMathematicsBusiness
221 pages, Kindle Edition
Published June 8, 2017
498 people are currently reading
146 people want to read
Ratings & Reviews
Friends & Following
Create a free account to discover what your friends think of this book!
Community Reviews
Displaying 1 - 12 of 12 reviews
May 12, 2020
Math is useful and in some sense beautiful. Nevertheless, there are thousands of bad teachers who can easily make you throw up while explaining mathematical matter.
Thanks to books like this, there is still hope that more people will appreciate the usefulness of numbers and start making conclusions in organised, logical way, void of bias and wishful thinking.
Thanks to books like this, there is still hope that more people will appreciate the usefulness of numbers and start making conclusions in organised, logical way, void of bias and wishful thinking.
January 12, 2021
Questo libro, in sintesi, prova a spiegare gli elementi delle formule che riguardano il coefficiente di determinazione, la correlazione, la regressione lineare, la regressione esponenziale, la regressione multipla.
Il metodo utilizzano non è particolarmente innovativo: con esempi e grafici vengono spiegati i principi di base. Ho trovato utile il link ad un foglio Excel contenente tutti i dati e i grafici esposti nel libro.
Ho trovato gli esempi un po’ troppo complicati e la parte finale sulla regressione multipla difficile da comprendere.
Appunti:
Coefficiente di determinazione (R2): è quanto la linea di regressione è meglio della semplice linea orizzontale che rappresenta il valore medio.
- 1.0 è il valore massimo
- 0 è uguale al valore medio
- < 0 è peggio del valore medio
Perché il Coefficiente di determinazione (R2) si basa sulla somma degli errori al quadrato medi?
Le alternative:
- somma errori (non al quadrato): gli errori con segno opposto si annullano.
- somma del valore assoluto degli errori (non al quadrato): non è differenziabile. La derivata nello 0 non esiste.
Correlazione vs regressione: la correlazione è un valore senza unità, mentre la pendenza della retta di regressione ha unità.
La correlazione è una misura di quanto strettamente due variabili si muovono insieme. Il coefficiente di correlazione di Pearson è una misura comune di correlazione e varia da +1 per due variabili perfettamente sincronizzate tra loro, a 0 quando non hanno correlazione, a -1 quando le due variabili si muovono l'una opposta all'altra.
Per la regressione lineare, un modo per calcolare la pendenza della retta di regressione utilizza la correlazione di Pearson.
È interessante notare che la correlazione zero non significa non avere alcun modello.
Nel caso di dati con andamento esponenziale, si utilizza il loro logaritmo per avere dati che hanno un andamento lineare.
La funzione di regressione di passa per la x e la y media. La funzione passerà per quel punto (x,y) e minimizzerà R al quadrato.
Ma possiamo scegliere di far passare la funzione per un altro punto, ad esempio il punto iniziale. Il punto scelto (x,y) verrà usato sia per b (la pendenza) che per a.
Il metodo utilizzano non è particolarmente innovativo: con esempi e grafici vengono spiegati i principi di base. Ho trovato utile il link ad un foglio Excel contenente tutti i dati e i grafici esposti nel libro.
Ho trovato gli esempi un po’ troppo complicati e la parte finale sulla regressione multipla difficile da comprendere.
Appunti:
Coefficiente di determinazione (R2): è quanto la linea di regressione è meglio della semplice linea orizzontale che rappresenta il valore medio.
- 1.0 è il valore massimo
- 0 è uguale al valore medio
- < 0 è peggio del valore medio
Perché il Coefficiente di determinazione (R2) si basa sulla somma degli errori al quadrato medi?
Le alternative:
- somma errori (non al quadrato): gli errori con segno opposto si annullano.
- somma del valore assoluto degli errori (non al quadrato): non è differenziabile. La derivata nello 0 non esiste.
Correlazione vs regressione: la correlazione è un valore senza unità, mentre la pendenza della retta di regressione ha unità.
La correlazione è una misura di quanto strettamente due variabili si muovono insieme. Il coefficiente di correlazione di Pearson è una misura comune di correlazione e varia da +1 per due variabili perfettamente sincronizzate tra loro, a 0 quando non hanno correlazione, a -1 quando le due variabili si muovono l'una opposta all'altra.
Per la regressione lineare, un modo per calcolare la pendenza della retta di regressione utilizza la correlazione di Pearson.
È interessante notare che la correlazione zero non significa non avere alcun modello.
Nel caso di dati con andamento esponenziale, si utilizza il loro logaritmo per avere dati che hanno un andamento lineare.
La funzione di regressione di passa per la x e la y media. La funzione passerà per quel punto (x,y) e minimizzerà R al quadrato.
Ma possiamo scegliere di far passare la funzione per un altro punto, ad esempio il punto iniziale. Il punto scelto (x,y) verrà usato sia per b (la pendenza) che per a.
December 22, 2024
It's a short book, simple and easy to understand, if you interest is a book most formal with mathematics and statistics rigorous this is not the book, but if you interest is something more easy to read or an introduction to regression, this book is ok because only includes a review necessary and also have example in excel where you could perfectly understand the idea of different topic that include.
January 2, 2019
Excellent Intro to regression
Very basic and simple yet effective introduction to regression. It is very simple but it delivers the material much better. Although it leaves out details, these book really gives a solid understanding of regression. Highly recommended for intro to econometrics students and practicioners.
Very basic and simple yet effective introduction to regression. It is very simple but it delivers the material much better. Although it leaves out details, these book really gives a solid understanding of regression. Highly recommended for intro to econometrics students and practicioners.
December 31, 2017
Clear and hands on
This is probably one of the best written book I have read explaining linear regression. I especially like how the author explains the multiple linear regression by using the simply linear regression steps.
This is probably one of the best written book I have read explaining linear regression. I especially like how the author explains the multiple linear regression by using the simply linear regression steps.
December 30, 2019
Too detailed. But topic is like that only
I liked the book, it has lot of maths, too much details but then only your can understand this topic in detail.
One who is using regression but want to understand all nuances, should definition pick this book.
I liked the book, it has lot of maths, too much details but then only your can understand this topic in detail.
One who is using regression but want to understand all nuances, should definition pick this book.
March 1, 2020
Very clear for beginners
I am learning several more advanced multi-linear regression classes right now, but I am sometimes confused due to lack of basic knowledge. This book is very clear in terms of basic concept of correlation, R square and regression.
I am learning several more advanced multi-linear regression classes right now, but I am sometimes confused due to lack of basic knowledge. This book is very clear in terms of basic concept of correlation, R square and regression.
December 13, 2018
Nicely explained
Author deliver what he promises. After reading this small book one will feel comfortable with regression and correlation. Recommend for beginners.
Author deliver what he promises. After reading this small book one will feel comfortable with regression and correlation. Recommend for beginners.
February 9, 2019
Enjoyable reading and very easy to follow.
I wish I'd had a teacher that teaches with the author's didactics when I was in school!
I wish I'd had a teacher that teaches with the author's didactics when I was in school!
June 29, 2019
Very good concise and explained well
Very good concise and explained well.
I recommend every budding statistics lover to read this. Hope physical copy is also available
Very good concise and explained well.
I recommend every budding statistics lover to read this. Hope physical copy is also available
August 10, 2019
Clear and short
February 17, 2020
Simple introduction to linear regression which tries to explain its algorithms in a somewhat visual way. Thought, would like to see more explanation on why it just works.
Displaying 1 - 12 of 12 reviews