Category: Real life Analytics


How can Neural Networks be used for forecasting?

Filed Under: Real life Analytics by analytic_atiblog — Leave a comment
May 10, 2011
Neural Network Image

The Simple logic behind Neural Network

The prediction of time series using neural network consists of teaching the net the history of the variable in a selected limited time and applying the taught information to the future. Data from past are provided to the inputs of neural network and we expect data from future from the outputs of the network.

An example of using neural networks for prediction: In today’s economy, most companies need to monitor their business expenses regularly and forecast accurately. They typically have a record of their business expense in the past. But most companies have multiple departments, and each of them have a different need of expenses. A general ARIMA time series model would require the company to forecast the expense for each of the department separately. Neural Network will help us to forecast the expense of company considering all the department and any other variables that are included.

Picture take from www.obitko.com

The prediction of time series using neural network consists of teaching the net the history of the variable in a selected limited time and applying the taught information to the future. Data from past are provided to the inputs of neural network and we expect data from future from the outputs of the network
Tags: ,
Comment

Segmentation – at our roadside!

Filed Under: Analytics, Real life Analytics by analytic_atiblog — Leave a comment
January 23, 2011

I was walking down the road after finishing my shopping from Westside and I stopped on the pavement to talk to my mom. A daughter-mom conversation typically takes more than 5 mins where the topics revolve anything from neighbours’ current status to the nosy relatives. And we were having one such conversation. And as I was talking over the phone, I happened to notice a beggar, who should be around 10-12 years of old, who made close to Rs.10/- in the first 2 mins of my conversation.

Then I could not but help recognize the pattern in the way she approached people. The girl approached only people with the following characteristics:

- Single men/women – when I say single, I do not mean unmarried but those that were by themselves – that were 20-32 years old

- All phirangs

- Those that did not travel on cars but were on foot

- Those that did not look like Delhities

And believe me or not, she was able to tap the best market by just concentrating on these segments. I was amazed that the girl who with no formal education was able to segment the market in the best way. And I, had to spend close to 2 years in a different land spending 30 lakhs to understand the concept of Segmentation!!

How many times have we questioned ourselves if the target segment chosen by us is the right? Well, in that case – determining the characteristics of the customers that they have and accordingly clustering it would do the charm.

Why did the girl choose this cluster?? More on this on the next blog…

Tags: , ,
Comment

Can a Model for Intuition be built?

Filed Under: Analytics, Real life Analytics by analytic_atiblog — Leave a comment
January 19, 2011

Going through all the possibilities of where Analytics could be applied was very interesting. Analytics could be applied in the field of sports, marketing, sales, retail, clinical, architecture, internet and the list can go on. I have come up with a very vague model  to define intution. I call this the modeling of gut or rather Modeling Intuition.

My younger brothers were preparing for JEE Exams and they had to practice answering loads of Multiple Choice Question’s(MCQ’s).That’s where the idea of this analytics model was born. I asked them to answer the paper 3 times.

  • The first time I asked them to answer the MCQ’s with proper solutions to the Questions, this would be the second variable, first variable being the question’s grouped in each subject(Math, Physics, Chemistry)
  • The second time I asked them to just answer the question’s with their intuition without actually using the pen and paper to solve the questions’s. this would dd as the 3rd variable.
  • The 3rd time I asked them to the solve the MCQ’s is a short span of time, which would be the 4th variable.
  • The correct answers would the 5th variable.

Modeling IntuitionAs of now these are the variables that I store. With simple analysis we could say that whether they would score above 70% intuitively or not, what would be the best practice that has to be followed using your intuition(we can confirm this with if one has constantly scored above 70% in at least in 50 MCQ papers). Another case we could measure is what percentage he has failed using intuition and how should he be acting . Obviously this is not a fool proof model and many other variables and factors should go in before we can actually start building a confidant model on Intuition. 

Modeling Intuition
Comment

Usage of DoE in our day-to-day activities!

Filed Under: Analytics, Design of Experiements, Real life Analytics by analytic_atiblog — Leave a comment
January 19, 2011

The first question that immediately props up is : “What is DoE?” . Well, the term may look fancy but it stands of Design of Experiment. Wikipedia may give a big gyaan on “How to model design of experiments” but let me tellyou something – you are already doing it without knowing.

Now, lets traverse through time a little. (Note: am talking about the time when there were only two channels on TV, Rajinikanth was paired with heroines of his age and the term ‘social media marketing’ was yet to be coined)

Analytics used by school girls to maximise marks

School Kid

When I was a kid, I used to appear to all my exams with a specific pencil (but not use it), start from home at 8:00 am and travel slowly – take about 25-30 minutes (when the exam was at 9:00 am and it takes less than 15 mins for me to travel), pass through a specific temple (well, I was god-fearing then), wear a blue coloured handkerchief (those were the days when tissue-papers were not known and mom used to pin a handkerchief to my shirt) and give exams. Well, the variables chosen by me may be dumb – but it worked! IT WORKED EVERY SINGLE TIME. Whenever I did not follow any of these or a combination of these, my marks have always gone down!

Coming to the analytics part, the dependent variable in this case is the marks scored and the independent variables are all the above ones in addition to the number of hours I studied. Now, design of experiments states that : a) Determine what is your end goal. In this case, to maximise marks with very less time spent on exams AND  b) Determine what are the dependent variables.

So, the formula in this case would be : Max marks = x * (# of hours studied) + y * (colour of handkerchief) + z * ( # of pencils carried) + a * ( # of minutes spent on travel)

I have tried changing every single time in changing one of the variables and seen what is the optimal # of hours that have to be spent on studying, what is average number of minutes I should spend on travel and how many pencils should I carry to maximise my marks.

So, now, I can give a formula to my success! :) Isnt analytics awesome?

Tags: , , ,
Comment
Powered by WordPress | Theme: Motion by 85ideas.