Skip to content

Latest commit

 

History

History
59 lines (42 loc) · 5.22 KB

README.md

File metadata and controls

59 lines (42 loc) · 5.22 KB
Raw

Books_I_Read

When we read a lot many books , sometimes you like some points in the book that you like to remember ,but after some point of time you tend to forget those valuable points, So this repo contains names of the books which i read and key points inside the book.

An Introduction to Statistical Learning (By - Gareth James, Daniela Witten, Trevor Hastie, Robert Tibshirani)

Chapter 1 (Statistical Learning.)

  • In statical Learning, we always try to find the association between variables, either by performing Bivariate analysis or by building the model and infereing from the model results.
  • To Get inference from the data, we always try to fit some perfect math function (f(x)) which has certain error(e) associated with it while predicting target variable(y) (y = f(x) + e, this error could be because of the confounding variables which effect target variable estimation.
  • Suppose we choosed certain math function (f_x) which we assum to be the perfect function (f(x)) and we try to fit and predict y_ for y independent variable, the accuracy dependents on two quantities 👎
    • Reducible error : In general f_x will not be the perfect estimate of f(x), this inaccuracy will introduce some error, this error is called Reducible error.We might decrease this error by just experimenting different math function on data and measure accuracy.
    • Irreducible error : The error occur due to unknow reason while infereing, this we can say Irreducible error.This error may occur due to confounding variables.
  • Inference:
    • We are ofter interested in understanding the way that target variable is effected as we change the predictors. In this situatoion we wish to estimate f(some function), but our goal is not necessarily to make predictions for target variable. We instead want to find the relationship between predictor and target variables, to be more specific, like how target variable change while we change predictiors. In this case f(x) can not be treated as black box.
    • Question we may as after modelling.
      • Which predictors are associated with target variable.
      • What is the relationship between predictor and target variables.
      • Can we summarize the relationship between target variable to each predictor variable by just linear functions.
  • Depending on whether our ultimate goal is prediction, inference, or a combination of the two, different methods for estimating f may be appropriate.
  • Ex :- Linear models allow for relatively allow for relatively simple and interpretable inference, but may not yield as accurate predictions as some other approaches.
  • Trade off Between Prediction Acuracy and model Interpretability.
    • We have two different types of models, in terms of flexibility of fitting to the given data.The less complex models are less flexible and more, if the model complexity is high, the its generally more flexible.
    • The less flexible models tend be more interpretable as compared to other flexible models.
    • The order of flexibility, 1 means less flexible, as the number increases the flexibilty of models decreases in the below order.
    • Sub set selection Lasso
    • Least Squares models
    • Generalized Additive Models.
    • Support vector Machines.
    • Bagging, Boosting.
  • Assessing Regression Model Accuracy.
    • Measuring the quality of fit (MSE)
    • MSE on training data :- The MSE on training data would be bit high for more flexible models, as the model flexibility decreases, the MSE may also decrease, as fitting of certain math function happens based on min square error statistic. But the major disadvantages of using more flexible models is, they tend to over fit due to this property.
    • MSE on Test data :- There will be small decrease in MSE of test data as compared to training MSE in more flexible models, but in less flexible models the MSE would be similar for both test and training in most of the cases, but it totally depends on data.
  • Bias-Variance Trade off.

YOU CAN WIN (BY :- SHIV KHERA)

chapter 1. (Importance of Attitude)

  • The greatest discovery of my generation is that human beings can alter their lives by altering their attitudes of mind.
  • When our attitude is right, we realize that we are all walking on acres of diamonds.Opportunity is always under our feet. We don't have to go anywhere. All we need to do is recognize it.
  • Opportunity only knocks once.The next one may be better or worse, but never the same one. That is why it is so crucial to make right decision at right time.
  • People with positive attitude have certain personality traits that are to recognize.They are caring, confident , patients and humble.
  • Our Behavior changes according to our experiences with various people.If we have a positive experience with a person, our attitude towards him will likely be positive.
  • Education referes to both formal and informal education. We are drowning in information but starving for knowledge and wisdom. Strategically applied, Knowledge translates into wisdom which in turn translates into success.
  • Take time to evaluate how the environment that you are in affects you, and the one you create effects you.
  • Total Quality people(TQP) are people with character, integrity, good values, and positive attitudes.

chapter 2.(How to build and maintain a positive attitude)