The importance of measuring Error

Most Reporting Systems tend to focus on the positive,

sales targetted ,sales achieved, etc.

For quality reasons , it is important to measure variance or deviation from the mean. This can also be interpreted as measuring error for any planning, forecasting or performance measurement exercise.

It helps bring focus, by aiming to reduce the difference between planned/forecasted and actual performance by looking at the mean absolute percentage error. Also it is impportant to have estimates that increasingly move from qualitative metrics or scale variables to quantitative inputs for forecasting or planning.

The forecasts should be realistic based on data. The forecasted pleasure or displeasure of the boss/ end audience recieving the estimate tends to be the biggest practical input in planning exercises unfortunately.

That is why analytics is sometimes a corporate cultural thing as well.

1) Forecast honestly

2) Use as much quantitative inputs

3) Plan scenario analysis with varying probabilties

4) Measure error from forecast and actual in a feedback loop

5) Aim at reducing error.

What SIX SIGMA would call

Define

Measure

Analyze

Improve

Control

.This applies to decision making and planning at most organizations where data /information ends up too many times in floating spreadsheets or gatekeepers/old timers.

Ballmer + Yahoo = Google's Boo Hoo

” I am going to fucking kill Google” Quote alleged to Steve Ballmer , 2005-6 ,due to Google’s poaching of engineers.

The following  scenario is completely a  work of fiction.kill-goo.JPG

11 Ways Ballmer and Yahoo can kill Google –

1)  Make their search engine better ,try combining the cache , and retrival queries at least for the first page, use some innovation like www.ask.com and www.guruji.com rather than pain vanilla search.. Maybe some more billions to buy out www.baidu.com —- But thats the toughest

2) Poach engineers from Google —- Thats hard

3) Combine their online ,offline advertising forces and spin it off in a different company. So there is one software company and one media company.

 4) Use Gmail’s emails feature to give more space and advertise based on email content in yahoo email and hotmail. Cross promote OfficeLive on Yahoo and vice versa.

5)PRICE-  Offer key words in a cheaper version of Adwords. Start offering an free alternative to Adsense AND SQUEEZE the long tail of Google’s revenue.

6)Use the facebook deal to sell more ads, give better ROI to advertisers, and monitor click fraud using Microsoft’s OS.

7) Buy out www.automattic.com for the wordpress and get a social community (facebook stake enhancement ??)

8) Offer desktop advertising as a freeware for some microsoft programs .

9)Tweak explorer to click repeatedly or not at all or at a slower speed on all ads that have the words ads by Google.

10) Make all Yahoo sites and Microsoft off search by Google, but searchable by yahoo.

11) Hire lawyers to repeatedly file anti trust cases against Google 🙂

Ballmer + Yahoo = Google’s Boo Hoo

” I am going to fucking kill Google” Quote alleged to Steve Ballmer , 2005-6 ,due to Google’s poaching of engineers.

The following  scenario is completely a  work of fiction.kill-goo.JPG

11 Ways Ballmer and Yahoo can kill Google –

1)  Make their search engine better ,try combining the cache , and retrival queries at least for the first page, use some innovation like www.ask.com and www.guruji.com rather than pain vanilla search.. Maybe some more billions to buy out www.baidu.com —- But thats the toughest

2) Poach engineers from Google —- Thats hard

3) Combine their online ,offline advertising forces and spin it off in a different company. So there is one software company and one media company.

 4) Use Gmail’s emails feature to give more space and advertise based on email content in yahoo email and hotmail. Cross promote OfficeLive on Yahoo and vice versa.

5)PRICE-  Offer key words in a cheaper version of Adwords. Start offering an free alternative to Adsense AND SQUEEZE the long tail of Google’s revenue.

6)Use the facebook deal to sell more ads, give better ROI to advertisers, and monitor click fraud using Microsoft’s OS.

7) Buy out www.automattic.com for the wordpress and get a social community (facebook stake enhancement ??)

8) Offer desktop advertising as a freeware for some microsoft programs .

9)Tweak explorer to click repeatedly or not at all or at a slower speed on all ads that have the words ads by Google.

10) Make all Yahoo sites and Microsoft off search by Google, but searchable by yahoo.

11) Hire lawyers to repeatedly file anti trust cases against Google 🙂

What is Analytics ?

Database mining and analytics are defined as using the power of hidden information locked in databases to reveal consumer and product insights, trends and patterns for future tactical and operational strategy. A key differentiator between analytics and market research is that analytics relies on data which is existing within a database while market research generally involves collection, collation and tabulation of the data.

Business intelligence is defined as the seamless dissemination of information throughout the organization and is a broader term, which involves and includes analytics as well as reporting systems.

The field of data analytics is vast and it comprises the following types.

Types of Analytics

  1. Reporting or Descriptive Analytics – Each organization relies on series on management information systems (commonly called MIS) to gather the current state of business as well as any emerging trend. This typically involves sales, finance, customer and competitor data, which is presented within spreadsheets and presentations. Reports tend to be either regular (like monthly and quarterly) or ad-hoc (for special investigative analysis). This is known as descriptive analytics simply because it describes the data which is present. While reporting or descriptive analytics is often the starting point in analytics careers, a proper grounding in this domain is necessary both to build an eye for detail in dealing with large amounts of data and for polishing the presentation skills for presenting insights from the data.

  2. Modeling or Predictive Analytics – Predictive Analytics refers to the art and science of using statistical tests, hypotheses and methods to build up predictive recommendations. These recommendations can range from which type of customer to call by phone for a credit card or insurance, to which type of mobile scheme to offer to a cell phone customer by a short message (sms) or to what kind of customers are likely to default on the loans they have taken. Predictive analytics includes techniques like segmentation, and regression modeling. It is generally considered both high value and a background in statistics helps in preparing for predictive analytics careers.

  3. Data-Driven Strategy – This is also called test –control or champion-challenger testing. This is done by segmenting the data population into test (on which a new strategy called the challenger strategy is to be tested) and control (which uses existing strategy called champion strategy). Building association rules which describe which parts of the product or customer data are clustered or co-related together are also part of analytics.

Basic Domains within Analytics

Data driven analytics by definition thrives in industries, which have large amounts of data, and high volume transactions, which need systematic and scientific analytics to cut costs and grow sales. The following domains offer employment opportunities to both new comers and experienced analytics professionals. These can be both in domestic firms, captive outsourcing firms or third party business process outsourcing companies.

  1. Retail Sales Analytics

Retail sales analytics deals with the handling of vast amounts of Point of Sales data, inventory data, payment data and promotional data to help increase sales in retail stores especially in organized retail. The use of RFID’s, Electronic Payment and Bar Scanning helps capture the data better and store it in vast databases. An example of this is the famous Thursday baby diaper-beer sales phenomenon. A big retailer found that on Thursday evening sales of beer and baby diapers were highly co related. It then found that is was due to young couples preparing for the weekend by buying supplies of diapers and beer. Thus by placing diapers and beer closer together sales could be boosted up. This is an example of market basket analytics in which a large amount of data is scrutinized to see which products sell well together. Wal-Mart, the American Retail Giant established a competitive edge over its rivals by proactively using data driven analytics to cut costs and thus offer goods cheaper than others. Another example of a big retailer is Target which has it’s own captive back end analytics in India.

In India, since Reliance Retail, Future Group, Walmart-Bharti have started setting up shop this is a sector that is bound to grow even within the domestic sector, as these high volume retailers need data driven decisions to squeeze the maximum from their retail stores.

  1. Financial Services Analytics

Financial services use analytics extensively. This is because they are in a very competitive field, have millions of customers and a lot of transactions. It is extremely important for them to store data for billing purposes and to recover the money they lend out as well the deposits they collect. An incremental gain of a few basis points (one hundredth of 1 percentage is a basis point) in profitability can lead to millions of dollars in aggregate profits. Within financial services analytics the broad sub categories are –

  1.  
    1. Risk and Credit Analytics – Risk and Credit functions measure the ability of a customer to pay back loans or debt owed by them. Delinquent customers are those that have fallen behind in paying back debt as per agreed schedule. Debt can be fixed installment like EMI’s for a personal loan, and debt can be revolving as in variable amounts that can be paid for credit card outstanding including the minimum balance. Debt can also be secured debt as secured against houses, consumer durables ,two wheelers, automobiles as collateral or it can be unsecured as in personal loans or credit card debt that have no collateral or backing. A risk analyst develops scorecards that help measure the risk worthiness of both new and existing customers. As financial service instruments are priced against risk, the riskier the customer the more they are charged in terms of interest rate. But this has to be balanced with total repaying ability of the customer, including sources of income and current leverage .In addition the income of customers especially in India is changing rapidly and there is also

un-declared income as black money. Doing the analysis for millions of customers is what make risk and credit analytics one of the hottest sectors to be in as credit analysts are in demand with all banks, outsourcing corporate and finance companies. ICICI has a big analytics unit (called Business Intelligence Unit) and Citigroup has both domestic analytics (in Chennai) and international analytics centers (in Bangalore).

  1.  
    1. Marketing Analytics- Marketing analytics helps in customer acquisition and retention. It does so by helping choose more responsive customers and selling through a wide variety of channels like call centers, direct mail, sms through mobiles, and email. It is marketing analytics which helps to bring in new customers by giving inputs to the marketing team and feedback to sales and distribution channels.

  1.  
    1. Collections –Collections Analytics focuses on recovery from delinquent customers using optimized efforts like telephones, direct mails, emails, or visits. Its aim is to maximize recovery at minimum costs.

    2. Fraud –Fraud Analytics seeks to build in triggers or automated alarms if there is any unusual trend or behavior in spending by the customer especially in credit cards.

    3. Pricing – Pricing Analytics tries to give the most optimized price, adequately compensating for risk as well as the competition. Pricing Analytics is a vast field and is also a part of financial services analytics especially in products like insurance.

  2. Telecommunications – Telecom Analytics has the fields of marketing analytics defined above, but an important part is also attrition modeling or churn analytics. It also analyzes the wide variety of pricing schemes and options and the customer response to them. In addition it has delinquency analytics as well.

  3. Pharmaceutical or Clinical Analytics –Clinical trials depend on test and control of thousands of patients on new drugs. Clinical trial analytics focuses on large number of variables that may or may not affect the drug response.

  4. Supply Chain Analytics – Supply chain analytics comprises inventory optimization, tracking turn-around time, multiple reports, and how to minimize the distribution costs.

  5. Transportation Analytics- Transportation analytics while covered more extensively in the field of operations research seeks at minimizing route length or fuel costs, or pricing of fares.

  6. Online or Website Analytics – Website analytics focuses on analyzing traffic to the website from sources, and how to retain them on the website for longer time or purchase more goods. It also involves a bit of search engine optimization to make sure the website is relevant in searches by search engines.

Vote for Monica Lewinsky

1) She has more experiences with Bill Clinton than Hillary.

2) She is younger , and a good first female President

3) Has seen tougher times than Hillary280px-billclintonmonicahug.jpg

4)She is also fresh enough to guarantee the politics of hope and “grope”

5) Has lesser negative approval ratings

6) She had bi partisan support in the 90’s

7) The economy was strong and HARD when she was in the OVAL Office.

8) Likely to be more pro gay, pro abortion (?), less reliant on lobbyist money

9) Her surname isnt Bush or Clinton. That makes a big change since 1988.

10) Al Gore and George Bush may just both endorse her.

http://www.famouspictures.org/index.php?title=Bill_and_Monica_Hug

Funny Story from College

This actually happened to me in Business School. At the end of year 1, we do summer internships. Some of the lucky ones then get job offers from the companies that they worked with in Summer.

students.JPG

Well I and two other batchmates worked with a company called G… So when G… came back on campus to take interviews from summer internships for the next year, they pre announced that they will give job offers or pre placement offers to 2 summer interns . We were thrilled. 2 out of 3.

Then comes my Placement Chairman, Professor M. He comes to me, shakes hands and congratulates me…saying hey Ajay, great job, You just got a job offer from company G… All my batchmates back slap me, some even give me job bumps .

2 hours later Professor M comes back, aplogizes and says he got the names wrong. I am the 1 out of the three who didnt get the job offer.

Leaving me awestruck, and my friends too embarassed to say anything.

Professor M, avoided me out of embarrassment ( I used to go out of my way to wish him Good Morning, Sir ) till end of second year , till I finally landed a job.

Ah ! College days!