[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
[Subscribe]
AW: DM: Datamining Definition.From: Frank Buckler Date: Tue, 4 Apr 2000 09:08:11 +0200 Defining DM , the Xth I feel not comfortable with all those definitions, so that's my try: For definitions we need objectives. In Data-Mining we apply some methods which should extract some insides out of data which relates to the objectives of a specific enterprise.... Maximizing Profits: To simplify business objectives to maximum profit, isn't a sensible definition from a managerial view. Profit is a short-term - term. Shareholder Value (SV) is a long-term one. But maximizing SV is a) a complex issue and b) do not guaranty a successful future. There are Stakeholder (e.g. moral of the public ... (look at doubleclick...) ) etc witch have to be considered ... So , this is just to complex for an definition. Lets better use instead of "maximizing profit" , "solve objectives which was derived from the management". Methods There are many scientific field from which we could derive some DM-Methods. I personally cant distinguish all the methods from there origin. This is also not our cup of tea. Let us describe the methods from there objectives too. What should this methods do? For me, the best descriptions of what DM-Methods do is "Learning", they examine inductive reasoning from Data. They do no confirmatory statistics and do no deductive reasoning. So here is my suggestion: DM is: "Acquiring knowledge (laws and regularities) automatically from data in order to achieve an business objective of an enterprise." Frank Buckler University Hanover Department Marketing II Buckler@m2.uni-hannover.de -----Ursprüngliche Nachricht----- Von: owner-datamine-l@nautilus-sys.com [mailto:owner-datamine-l@nautilus-sys.com]Im Auftrag von BHALERAO, Narayan Gesendet: Montag, 3. April 2000 07:08 An: datamine-l@nautilus-sys.com Betreff: RE: DM: Datamining Definition. Hello, I tend to agree with Werner. His definition is closer to my definition. However, I disagree that you need large and messy data sets to do data mining. The main purpose of data mining is to maximize profit and minimize risks. The methodology is immaterial whether it is based on Statistics, OR, AI/ML or any other technique. The methodology could be formal or informal. The main goal is to be valuable to the business. Data mining goes much further than just theoretical methodology and includes driving towards the main objective of maximizing profits and minimizing risks. I, therefore, define data mining as a formal/informal process of extracting "valuable information" from the data sets to maximize profit and minimize risks. In most cases these data sets are large and messy. Narayan Bhalerao Data Mining Manager Westpac Banking Corporation Level 6, 341 George Street Sydney 2000 NSW Australia Phone: 61-2-9220-3706 E Mail: nbhalerao@westpac.com.au > ---------- > From: Werner E. HELM[SMTP:helm@fh-darmstadt.de] > Reply To: datamine-l@nautilus-sys.com > Sent: Friday, 31 March 2000 21:33 > To: datamine-l@nautilus-sys.com > Subject: Re: DM: Datamining Definition. > > > Hi all : > > Warren Sarle wrote : ..... > > So I choose to define data mining as the application of statistical > decision theory to huge, messy data sets to maximize profits. > > Well, coming myself from a background of Statistics and Operations > Research, I could agree. > However, it depends on how wide or narrow you understand "statistics". > > A bit wider would be : > > " So I choose to define data mining as the application of any formal > method > around > decision theory to huge, messy data sets to maximize profits." > > Formal method would comprise Statistics, Operations Research,AI/ML-methods > > and maybe future concepts from Mathematics. > > Since I very seldom have experienced that the results of decision theory > go > to decisions without human interaction of some kind of management guy, I > could return to "decision support" and suggest : > > " So I choose to define data mining as any type of decision support in all > > cases of huge, messy data sets to maximize profits." > > Of course decision support methods would mainly come from Statistics, > Operations Research,AI/ML-methods and maybe future concepts from > Mathematics. > > > But : A sales argument needs no definition. > But : A strong movement will find it's way on campus, attached to > CS-dept., Stats-dept., OR-dept. or between existing disciplines. If > strong > and persistent enough it will create an own existence. > > So let's wait and see how strong and persistent DM will be when the > sales > people move on !! > > > Werner E. HELM . >
|
MHonArc
2.2.0