modeling · non-fiction · review

Predictive analytics- The power to predict who will click,buy,lie,or die.Eric Siegel

Predictive Analytics book cover 3D - thumbnail

Oh yeah me and my blogs. Do they really sound too personal-Something that shouldn’t be shared with public. Would people really bother to read my blogs or they don’t bother. Or they might read them when I’m gone forever or would they be used for text analytics as one among the crowded blogging population. These are some of the thoughts that triggered when I started writing this one. Okay, the reason why so many thoughts got triggered was My tummy full of bitter chocolate ice cream fudge that was a treat from one of my friends.I need to thank Elizabeth Gilbert for throwing a light on my mind that one can write anything.

Statistics– This word became a fad to me in early 2013 when one of the program managers in the client location shared the link of udacity to everyone other than me. He might have thought that it is not worth to educate me on udacity since he felt I was not worth it and my knowledge is below the expectation. Sad, but true, that gentleman doesn’t know that I had 100% score in business statistics during my college days and far more mature in my early days of career in hp and I was one of the folks who got promoted four times in five years of career in hp and exceeded all the expectations in my company and thanks to my ex-boss for god-fathering me then. Actually, I did not care of the program manager’s intentions, coz I peeked into my neighbor’s computer and noted the website down. Sometimes people’s abstinence of you from doing stuff will make you crave more of what they don’t want you to do or do to get there. Yes, I got there. I got closer and I’m trying hard to excel, lead and win. I will be the apple-pie one fine day and the key ingredient in the platter of the company .I will become what I want if not Mark Zuckerberg or Sheryl  Sandberg.

So coming back to the story of 2013, I had the hunger to enter into big data and started brushing up algebra .Then I came across Eric Siegel’s book on Predictive analytics- The power to predict who will click, buy , lie, or die. My brother-in-law suggested me to read this book. Instead of borrowing the book, I purchased this for myself, placed an order with the infibeam.com. Curiosity struck when I received the package. I had unwrapped the package and found the book, hidden inside the box. I was really very interested and stopped reading the Harry potter and the deathly hallows and started reading this. The adrenaline rush and boom got super stuck. Couldn’t understand a hook or crook. Sadness spread like an ocean and I had started weighing the return on investment of this book. This was supposed to be one of the most expensive books hard bound, imported edition and there my conscience started to prick. Do I really need to return this book and then my confidence was rekindled by my mom who had asked me to retain this book and keep it for future reference.

Then came the year 2014- The year I roamed on the roads of Hyderabad to learn analytics, borrowed huge amount of money from my hubby and paid an extremely handsome amount to the institute that taught me what analytics is and with their fake promises of the job offers. I had to relocate to Chennai due to the personal reasons and the book went back to the shelves and was one among the huge list of fiction I had.

The year 2015- This year is of my life is called Being idle and the year that increased the money I owe to my husband. It was Rs.50000 in 2014 which became Rs.500000 in this year. 10X fold debts. Apparently, this year also seemed to be one of the most productive in terms of improving my skill set on analytics, communication, pursuing hobbies, reading, writing blogs, cooking, etc.  I paid a visit to my mom’s for a month. There I happened to find this book written by Eric Siegel and started reading the book. My first impression of this one for the second time was that I should read this one before I travel back to Texas.  Hence, I had started the journey of reading this book. I was able to relate and understand the concepts better this time since I had started doing few miniscule analytics projects at the beginner’s level.

The book is divided into seven chapters. Each chapter explains one concept used in machine learning/data analytics. The chapters have real world examples, that provide a strong foundation on the concept .Apart from the examples, this also has anecdotes from various authors that is relevant to the concepts explained therein. The cartoons mentioned in some of the chapters make the reader understand the logic with less effort. The back and front flaps of the book has an introduction of the author and the praises for his work. Each application of predictive analytics(PA) is defined by what’s predicted and what’s done about it. This book also has 147 examples of predictive analytics across various industries right from family and personal life, financial risk and insurance, health care, crime fighting, government, psychology and human resources. This book concentrates on five effects-The prediction effect, The data effect, The induction effect, The ensemble effect and the Persuasion effect. He also explains that in order to choose the title of the book, he had to run an online competition based on clicks of the users. The title that had maximum clicks was chosen to be the title of the book.

The introduction of the book runs for few pages that talk about what is big data, predictive analytics, machine learning and also provides a brief summary of what has been explained in the forthcoming chapters of the book. He also talks about the limitations of the prediction. According to the author, predictive analytics is a technology that learns from experience (data) to predict the future behavior of individuals in order to drive better decisions.

The chapter one, that is named as ‘Liftoff! Prediction takes Action’ explains about Dr.John Elder who had devised his own predictive model for stock trading and the success rate he had achieved. He had invested almost all retirement money in stock trading and took a chance to experiment his model which was a success. This chapter also explains about the PA application in the target direct marketing, predictive advertisement targeting and black box trading. The chapter two, that is named as ‘With Power Comes Responsibility’ talks about the pregnancy prediction of Target, crime prediction of a future crime in a location and the PA application of employee retention. The chapter three, that is named as ‘The Data Effect’. This talks about the relationship between human behavior and emotions. I was personally touched by this chapter when Eric explains how the moods in the blogs were related to the stock market which is measured by a parameter called anxiety index. The next chapter is named as ‘ The machine that learns’ which insights about Chase bank’s prediction of mortgage risk. This chapter talks about how the decision trees as a tool used to predict the mortgage risk. The next one The ensemble effect talks about stuff in brief crowdsourcing, super charging prediction and what Netflix did to recommend movies to the customers at large. The ensemble modeling is a bag of models which has a key characteristic of diversity such as CART, random forests, logistic regression, etc. The next one talks about how IBM had created Watson that was able to challenge the human brains and what technique they had used to train their machine to beat the human brains. Rather than just a search that would get the responses based on keywords, Watson had used a technique of predictive modeling based on the training dataset collected from the fans of the Jeopardy show on the television. Ensemble and logistic regression were used. This chapter was named as ‘Watson and the Jeopardy!Challenge’. The final chapter ‘Persuasion by the Numbers’ explains how Telenor had reduced the attrition rate of their customers by churn rate. Rather than concentrating on how many customers respond to the brochures sent, the emphasis was on to target the persuadable customers. This is called uplift modeling.The uplift model takes into account the treatment and control dataset to arrive at the best model.He also explains as to how persuasion by the numbers influencing several applications across the industries.His final example was: how Obama ‘s analytics team managed to get data to campaign the voters according to the preferences of an individual. Who needs to be contacted through mail, in-person, digital, call, flyer, door knock or TV ad and avoid the ones that would be adversely influenced if contacted.

The afterword of the book has ‘Ten predictions for the First Hour of 2020’.According to Eric on January 2nd 2020, the first workday of the year, predictive analytics would be involved in predicting the following Antitheft, Entertainment , Traffic, Breakfast, Social,Deals, Internet search,Driver inattention,collision avoidance and reliability.

This book comprises of three appendices the five  effects, the twenty-one applications of PA prediction people-Cast of “Characters” along with notes talking about how he brought it the live examples and the sources of the study.

To conclude on this review,apart from newcomers, even folks who have an experience can read this book so that, it will give them a hand-hold as to how the concepts of PA and machine learning can be applied in a real-world situation. I have given a rating of four out of five in good reads for this book because of its very good explanation and the book had bridged the gap between theory and practice. Why four and not five because, I got this right only in my second attempt to read this book.

Advertisements

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s