possible data mining applications

neilmpatel

So, I am new to the world of neural nets and predictive intelligence and wanted to know what many of you on these forums thought about various applications of rapidminer. I'm writing a report on emerging technologies and wanted to list out all the various applications of business intelligence software like rapidminer.
so far, i have the following:
financial lending institutes predicting credit scores for individuals
preventive maintenance for machines
classifications of wines or products based on chemical compositions (this was in the rapid miner webinar)
classification of general health outcomes based on underlying data
financial asset price predictions
text mining and frequent item set (the popular one is the grocery store example but i suppose the vancouver data example of job posting classification is nice as well)
decision tree and market segmentation (this can be in virtually any business with customers).

can anyone think of any other business applications for neural networks or some of the operators in rapid miner?

shikhar

could it be used in logistics applications

MariusHelf

Sure, data mining, business intelligence and business analytics can be used in virtually any field. Btw, data mining is more than neural nets; in fact, neural nets are one of the more ugly parts and there are much more elegant solutions which perform the same or better task as neural nets.

However, besides other fields, our customers apply our products for:
marketing analysis, yield prediction, fraud detection, preventive machine/production surveillance, analysis of physical and astronomical experiments, and a whole bunch of other fields.

And there is surely potential for the use of data mining in logistics, just tell us a bit more what you are thinking about

Regards,
  Marius

shikhar

kk for starters lets say we have a data containing origins,destination,distance of shipment ,the volume of shipment and the value of shipment from all 50 states to all 50 states in USA from 1990 till 2010.Can you tell me any possible application of data mining on the data specifically ??

MariusHelf

Let me answer with an allegory: if you go to the carpenter, put a couple of planks onto his table and ask him "can you construct something with this?", he will surely say yes, because every carpenter can construct a whole lotta things out of a pile of planks. But if you want something which is useful for you, you need to give him more information: "I need something where I can put my plates/where I can sleep/where my children can play/...".
Same for data mining: we can apply some kind of datamining on the data. But without knowing what information you are seeking, it will probably not make much sense.

Anyway, this is a selection of things RapidMiner can do for you with your dataset:
- graphically present the development of any value in the data, e.g. shipping value per state
- predict the development of of shippings from one state to another (provided the fact that all influence factors for shipping development is in the data)
- predict the origin of the shipping, given volume, destination, distance and date
- ...

Best,
  Marius

shikhar

Is it possible that with the kind of data that i have given in the previous description i can predict the demand of shipment from one particular origin say alabama to various other 49 destinations and also find any similarity in terms of price per unit tonne transported in different routes

dudester

I'm also a newbie, and I don't know much about the transportation world, but I do know, for instance, in education teachers/school districts are constantly trying new curricula and testing for outcomes.  Data mining is logical here in that it can isolate specific factors, depict their relationships visually, and give clues as to underlying meaning. I may see that gender impacts career choice greater than another particular intervention, for example.  In some cases it can "flesh out" relationships that traditional statistics either ignores or cannot model well.  In fact, you can/should use DM to help build models to mimic the real world understanding of what's happening, so you might introduce improvements you can test.  Content area knowledge is key- sounds like you know the transport world.  Learn about model building, and use the content area knowledge you already have.