Yelp And Angie's List: The Paradox Of The Local Review Listing Service [View article]
ANGI, like GRPN postulates 'stickiness' of customers...they point out 70% of of first year customers renew.
Nonetheless, other issues remain. For instance, ANGI's main focus area is services for the home. Now if you are dissatisfied with the contractor, would you write a negative review, knowing fully well that any follow up repair/warranty service will come from a hostile contractor?
Secondly, I know this from personal experience, many restaurants and local businesses basically write fake reviews about themselves and the competition and the fraud algorithms aren't really able to spot it. The quality of the reviews is inherently suspect.
Beware The Hype Over Big Data Analytics [View article]
Hello clf28264
I hear you on accounting!! Yes, to do a good model design takes years of experience to do the model/dataset/feedback loop, years during which one can understand the field. A consultant can't provide it as they will come and go and yet, as you yourself point out, during this time it is hard to quantify the return to the bottom line and pay a statistician for years.
As for unstructured data, without good metadata and lacking data integrity checks, you can perhaps extracts insights, but usually not decisions. Organizationally someone have to be the champion of decisions based on data that has a lot of error and has no meta data. You'll find yourself on the other side of the table with someone who can wield the same data to draw different conclusions.
I definitely think the very good statisticians (i.e. the Ph.D.'s from MIT type) can add value, but these are a rare species, not easily found or kept. For the most part, the "analytics" sold by the firms with "software factories" in Bangalore, that's the stuff I mentioned could be done with Excel/VBA. I'm not suggesting that that is the best way to do statistics, but that if everyone was trained in Excel/VBA compared to nothing, their productivity would shoot up.
My thrust is a) for the simpler work (80%), the average manager and executive can do it themselves with rapidly more user friendly stat tools b) for the more complex work (20%) you need people with both deep stat skills and industry knowledge.
So for neither of the above are the data analytics firms needed - hence I don't see the hype over them.
Big Data To Grow 500% By 2015 And 2 Companies Are Uniquely Positioned To Benefit [View article]
More data is not linearly equal to more value for companies themselves. Might be profitable *in the short term* for the hardware and software providers. The value of "big data" is more of a exponential converging curve. My contrary take (for those who are interested in such matters) here -
Beware The Hype Over Big Data Analytics [View article]
Hello rad_123,
Thanks for your comments. I'll address what I can...
"your article missed other industries and as a news flash, they are needing ways to extrapolate information to improve their operations. "
I have no doubt. But I'm unable to find out which - not from the popular media which has a fixation on retail/online sales and credit card/credit score analytics. Not on the surface from the websites of the analytic firms themselves. Remember, I'm looking for industries that are new to data analytics and/or data collection.
"new industries, government, medical, and institutions" Certainly the government could use help analyzing all the data they collect about the people. Whether that is desirable is another matter. In the medical field, the electronic medical records (EMR) is scattered among 30 vendors. Over in Britian the effort to develop a single EMR by Computer Sciences Corp. has ended in a crashing failure. Part of Obamacare's push to force medical practices to use EMR was to try to analyze this data to determine how to reduce treatment and expense via the independent payment advisory board. It's a tall order and probably doomed to failure and certainly doomed to undying hatred of the citizenry like the NICE board of Britain.
"having the ability to calculate such equations is complex and could become convoluted. "
I would suggest there is some really complicated stuff and mostly pedestrian stuff in the usual 80:20 ratio. For the 80%, take a very simple consumer level program like Minitab. Within Minitab, you can do multivariate regression, step regression, binary logistic regression, various ANOVA, factor analysis, etc etc with minimal training. With 8-16 hours of training, 75% of middle managment (who have some math courses in college) can be trained to use these tools to at least a usable extent. For STEM graduates this a guarantee - this is how analytics consultants are produced after all. For the remaining 20% yes, probably you do need a Ph.D. to model the airflow over a airfoil via finite element analysis and design a better one. But these sort of people don't come from analytics firms! You hire them, develop them and pay them to stay with you.
"the version X only imposed 2 faults in the production due to the design but version XXXI imposed 15 faults."
Yes, you are right. We use all sorts of analytics in design and operation of medical devices both at device level and from a aggregate level. At device level, it might be to determine a true even from a false positive or to seperate the signal from the noise. At the aggregate level, we may use data from many devices and many events and process this data to determine various risk scores or disease progression indices. But this a small part of our costs. Mostly the expenses are in distribution, marketing and manufacturing, with a much smaller component in R&D. Our distribution costs in 1 year would pay for 20 years of the spend on data analytics. If a firm wanted to make money off of us, they would do far better to be distributor than a data analytics vendor! I would assume this to be true for most (though not all industries).
Apple: All-Glass iPhone 5 Draws One Step Closer [View article]
The schema depicts common process steps for any glass coated product. Remember you can't patent the application, only the elements.
If this ridiculous patent is granted, what's next? A patent for using printed circuit boards? Both in Germany and US, Apple's patents have failed to hold up when challenged. Judges have raised concerns about the validity of the issued patents. It may give "ammo" to Apple, but mainly it'll be the kind of ammo that blows up in their face.
Apple hasn't won a single significant victory. The few they managed to get in Europe have all been overturned. They've won nothing in the US. The patent office will basically issue a patent for a wheel, if the patent is suitably worded. In the medical device industry, there are so many identical patents, so many contradictory patents, it's clear that anyone can get a patent for anything. Given the recent Supreme court ruling making it easier to challenge issued patents in court at a later date, I anticipate that 95% of Apple's patents will not hold water.
I would not read too much "moat" potential in Apple's patents. I've found the few patents that have made it to the news to be remarkably weak. For some strange reason, they have made a big part of their business to support the livelihood of trial and IP lawyers by engaging in futile lawsuits that bring them nothing.
Beware The Hype Over Big Data Analytics [View article]
Hi K_canada,
I wonder if even inhouse teams are needed? Would it not be possible in the future for an average manager/executive to do their analysis on their own? Though data analysis is not exactly typing a document, I keep thinking of Microsoft Word and the typing pool. A really smart data analyst is of course on a couple of magnitudes of skill level above a typist. But still if the person with the industry knowledge can do their own analysis, that should be the goal (sort of like the applications tailored for users that you refer to, to do their own analysis)
Beware The Hype Over Big Data Analytics [View article]
I agree with your point. I think what I was trying to say is that the middleman is rapidly becoming unnecessary. If you are Salesforce.com or Amazon, if you develop a tool, you can as well offer it to all your customers. Whereas when a large business develops a analytic tool, they guard it fiercely as a asset.
I think smaller players will be not smaller firms specializing in data analytics, since they'll still be unaffordable for most small and medium businesses. But the smaller players probably will be the actual businesses themselves, doing the analytics using the tools either provided by the cloud providers or 3rd party such as SAS etc.
Beware The Hype Over Big Data Analytics [View article]
The cloud thing essentially seems a reprise of the whole mainframe idea except that instead of one mainframe serving one facility or company, you have lots of little mainframes wired together serving lots of companies.
Beware The Hype Over Big Data Analytics [View article]
To clarify, what I mean is small and medium business can acquire the tools to do the work themselves cutting out the middle man of data analytics firms and large businesses are acquiring the smaller providers rather than pay $100-200/hour.
Beware The Hype Over Big Data Analytics [View article]
If companies would spend 3 days a year training people in Excel and basic VBA (so that they can write thier own formulas and functions), you would see a huge increase in productivity.
Interestingly the most useful and simple tool that many many people have told me about and that I also found to be superb is a old product called RS1 (I haven't seen it for many years and I don't believe it is available any more) that allowed absolute beginners to use almost natural language terms to query databases. Even today, I believe it is superior to many of the tools out there.
Beware The Hype Over Big Data Analytics [View article]
Yes, I too am waiting for the CRM dam to break. However, it may be acquired by some deep pocketed company (HP comes to mind) for a even bigger premium. Hence, shorting CRM is not a wise idea.
Beware The Hype Over Big Data Analytics [View article]
Hi Ernests,
Yes, the lowering in price and increase in user friendliness of data collection and analysis effectively gives access to small and medium business who cannot afford external help. But also for big business, it helps them develop inhouse teams. For example large companies like United Healthcare have been on an acquisiton spree, buying up smaller data analytics firms and converting them to inhouse consultants.
Yelp And Angie's List: The Paradox Of The Local Review Listing Service [View article]
Nonetheless, other issues remain. For instance, ANGI's main focus area is services for the home. Now if you are dissatisfied with the contractor, would you write a negative review, knowing fully well that any follow up repair/warranty service will come from a hostile contractor?
Secondly, I know this from personal experience, many restaurants and local businesses basically write fake reviews about themselves and the competition and the fraud algorithms aren't really able to spot it. The quality of the reviews is inherently suspect.
How Goldman Sachs Stole Silicon Valley [View article]
Learn To Embrace Losing And You Will Make More Money [View article]
As spring blossoms bloom
A cool million readers
Watch their stocks grow too.
Enter our Seeking Alpha 1 Million User Haiku Contest, by Tuesday at midnight, and a new iPad could be yours! (earlier) [View news story]
The story of my money
Is a happy tale
Beware The Hype Over Big Data Analytics [View article]
I hear you on accounting!! Yes, to do a good model design takes years of experience to do the model/dataset/feedback loop, years during which one can understand the field. A consultant can't provide it as they will come and go and yet, as you yourself point out, during this time it is hard to quantify the return to the bottom line and pay a statistician for years.
As for unstructured data, without good metadata and lacking data integrity checks, you can perhaps extracts insights, but usually not decisions. Organizationally someone have to be the champion of decisions based on data that has a lot of error and has no meta data. You'll find yourself on the other side of the table with someone who can wield the same data to draw different conclusions.
I definitely think the very good statisticians (i.e. the Ph.D.'s from MIT type) can add value, but these are a rare species, not easily found or kept. For the most part, the "analytics" sold by the firms with "software factories" in Bangalore, that's the stuff I mentioned could be done with Excel/VBA. I'm not suggesting that that is the best way to do statistics, but that if everyone was trained in Excel/VBA compared to nothing, their productivity would shoot up.
My thrust is a) for the simpler work (80%), the average manager and executive can do it themselves with rapidly more user friendly stat tools b) for the more complex work (20%) you need people with both deep stat skills and industry knowledge.
So for neither of the above are the data analytics firms needed - hence I don't see the hype over them.
Big Data To Grow 500% By 2015 And 2 Companies Are Uniquely Positioned To Benefit [View article]
http://bit.ly/GUcLiN
Beware The Hype Over Big Data Analytics [View article]
Thanks for your comments. I'll address what I can...
"your article missed other industries and as a news flash, they are needing ways to extrapolate information to improve their operations. "
I have no doubt. But I'm unable to find out which - not from the popular media which has a fixation on retail/online sales and credit card/credit score analytics. Not on the surface from the websites of the analytic firms themselves. Remember, I'm looking for industries that are new to data analytics and/or data collection.
"new industries, government, medical, and institutions" Certainly the government could use help analyzing all the data they collect about the people. Whether that is desirable is another matter. In the medical field, the electronic medical records (EMR) is scattered among 30 vendors. Over in Britian the effort to develop a single EMR by Computer Sciences Corp. has ended in a crashing failure. Part of Obamacare's push to force medical practices to use EMR was to try to analyze this data to determine how to reduce treatment and expense via the independent payment advisory board. It's a tall order and probably doomed to failure and certainly doomed to undying hatred of the citizenry like the NICE board of Britain.
"having the ability to calculate such equations is complex and could become convoluted. "
I would suggest there is some really complicated stuff and mostly pedestrian stuff in the usual 80:20 ratio. For the 80%, take a very simple consumer level program like Minitab. Within Minitab, you can do multivariate regression, step regression, binary logistic regression, various ANOVA, factor analysis, etc etc with minimal training. With 8-16 hours of training, 75% of middle managment (who have some math courses in college) can be trained to use these tools to at least a usable extent. For STEM graduates this a guarantee - this is how analytics consultants are produced after all. For the remaining 20% yes, probably you do need a Ph.D. to model the airflow over a airfoil via finite element analysis and design a better one. But these sort of people don't come from analytics firms! You hire them, develop them and pay them to stay with you.
"the version X only imposed 2 faults in the production due to the design but version XXXI imposed 15 faults."
Yes, you are right. We use all sorts of analytics in design and operation of medical devices both at device level and from a aggregate level. At device level, it might be to determine a true even from a false positive or to seperate the signal from the noise. At the aggregate level, we may use data from many devices and many events and process this data to determine various risk scores or disease progression indices. But this a small part of our costs. Mostly the expenses are in distribution, marketing and manufacturing, with a much smaller component in R&D. Our distribution costs in 1 year would pay for 20 years of the spend on data analytics. If a firm wanted to make money off of us, they would do far better to be distributor than a data analytics vendor! I would assume this to be true for most (though not all industries).
Apple: All-Glass iPhone 5 Draws One Step Closer [View article]
If this ridiculous patent is granted, what's next? A patent for using printed circuit boards? Both in Germany and US, Apple's patents have failed to hold up when challenged. Judges have raised concerns about the validity of the issued patents. It may give "ammo" to Apple, but mainly it'll be the kind of ammo that blows up in their face.
Apple hasn't won a single significant victory. The few they managed to get in Europe have all been overturned. They've won nothing in the US. The patent office will basically issue a patent for a wheel, if the patent is suitably worded. In the medical device industry, there are so many identical patents, so many contradictory patents, it's clear that anyone can get a patent for anything. Given the recent Supreme court ruling making it easier to challenge issued patents in court at a later date, I anticipate that 95% of Apple's patents will not hold water.
I would not read too much "moat" potential in Apple's patents. I've found the few patents that have made it to the news to be remarkably weak. For some strange reason, they have made a big part of their business to support the livelihood of trial and IP lawyers by engaging in futile lawsuits that bring them nothing.
Beware The Hype Over Big Data Analytics [View article]
I wonder if even inhouse teams are needed? Would it not be possible in the future for an average manager/executive to do their analysis on their own? Though data analysis is not exactly typing a document, I keep thinking of Microsoft Word and the typing pool. A really smart data analyst is of course on a couple of magnitudes of skill level above a typist. But still if the person with the industry knowledge can do their own analysis, that should be the goal (sort of like the applications tailored for users that you refer to, to do their own analysis)
Beware The Hype Over Big Data Analytics [View article]
I think smaller players will be not smaller firms specializing in data analytics, since they'll still be unaffordable for most small and medium businesses. But the smaller players probably will be the actual businesses themselves, doing the analytics using the tools either provided by the cloud providers or 3rd party such as SAS etc.
Beware The Hype Over Big Data Analytics [View article]
Beware The Hype Over Big Data Analytics [View article]
Beware The Hype Over Big Data Analytics [View article]
Interestingly the most useful and simple tool that many many people have told me about and that I also found to be superb is a old product called RS1 (I haven't seen it for many years and I don't believe it is available any more) that allowed absolute beginners to use almost natural language terms to query databases. Even today, I believe it is superior to many of the tools out there.
Beware The Hype Over Big Data Analytics [View article]
Beware The Hype Over Big Data Analytics [View article]
Yes, the lowering in price and increase in user friendliness of data collection and analysis effectively gives access to small and medium business who cannot afford external help. But also for big business, it helps them develop inhouse teams. For example large companies like United Healthcare have been on an acquisiton spree, buying up smaller data analytics firms and converting them to inhouse consultants.