Information Mining Techniques for CRM .

Uploaded on:
Category: Product / Service
Data Mining Techniques for CRM. Seyyed Jamaleddin Pishvayi Customer Relationship Management Instructor : Dr. Taghiyare Tehran University Spring 1383. Outlines. What is Data Mining? Data Mining Motivation Data Mining Applications Applications of Data Mining in CRM Data Mining Taxonomy
Slide 1

Information Mining Techniques for CRM Seyyed Jamaleddin Pishvayi Customer Relationship Management Instructor : Dr. Taghiyare Tehran University Spring 1383

Slide 2

Outlines What is Data Mining? Information Mining Motivation Data Mining Applications of Data Mining in CRM Data Mining Taxonomy Data Mining Techniques

Slide 3

Data Mining The non-trifling extraction of novel, verifiable, and significant learning from substantial datasets. To a great degree extensive datasets Discovery of the non-evident Useful information that can enhance procedures Can not be done physically Technology to empower information investigation, information examination, and information perception of expansive databases at an abnormal state of deliberation, without a particular speculation as a main priority . Refined information look ability that utilizations measurable calculations to find examples and connections in information.

Slide 4

Data Mining (cont.)

Slide 5

Data Mining (cont.) Data Mining is a stage of Knowledge Discovery in Databases ( KDD ) Process Data Warehousing Data Selection Data Preprocessing Data Transformation Data Mining Interpretation/Evaluation Data Mining is now and again alluded to as KDD and DM and KDD have a tendency to be utilized as equivalent words

Slide 6

Data Mining Evaluation

Slide 7

Data Mining is Not … Data warehousing SQL/Ad Hoc Queries/Reporting Software Agents Online Analytical Processing (OLAP) Data Visualization

Slide 8

Data Mining Motivation Changes in the Business Environment Customers turning out to be all the more requesting Markets are immersed Databases today are colossal: More than 1,000,000 substances/records/lines From 10 to 10,000 fields/characteristics/factors Gigabytes and terabytes Databases a developing at a remarkable rate Decisions must be settled on quickly Decisions must be made with most extreme information

Slide 9

PHOTO: LUCINDA DOUGLAS-MENZIES PHOTO: HULTON-DEUTSCH COLL Data Mining Motivation "The key in business is to know something that no one else knows." —Aristotle Onassis "To comprehend is to see designs." —Sir Isaiah Berlin

Slide 10

Data Mining Applications

Slide 11

Data Mining Applications: Retail Performing wicker bin investigation Which things clients tend to buy together. This information can enhance stocking, store design methodologies, and advancements. Deals estimating Examining time-based examples helps retailers settle on stocking choices. In the event that a client buys a thing today, when are they liable to buy a correlative thing? Database showcasing Retailers can create profiles of clients with specific practices, for instance, the individuals who buy creator marks attire or the individuals who go to deals. This data can be utilized to center cost–effective advancements. Stock arranging and allotment When retailers include new stores, they can enhance stock arranging and distribution by looking at examples in stores with comparable demographic attributes. Retailers can likewise utilize information mining to decide the perfect format for a particular store.

Slide 12

Data Mining Applications: Banking Card promoting By distinguishing client portions, card backers and acquirers can enhance benefit with more successful obtaining and maintenance programs, focused on item advancement, and altered valuing. Cardholder estimating and gainfulness Card backers can exploit information mining innovation to value their items to boost benefit and minimize loss of clients. Incorporates hazard based estimating. Misrepresentation recognition Fraud is immensely expensive. By dissecting past exchanges that were later resolved to be deceitful, banks can recognize designs. Prescient life-cycle administration DM helps banks anticipate every client\'s lifetime esteem and to benefit every section suitably (for instance, offering unique arrangements and rebates).

Slide 13

Data Mining Applications: Telecommunication Call detail record examination Telecommunication organizations amass itemized call records. By recognizing client fragments with comparable utilize designs, the organizations can create appealing evaluating and highlight advancements. Client faithfulness Some clients more than once switch suppliers, or " stir ", to exploit appealing motivating forces by contending organizations. The organizations can utilize DM to recognize the attributes of clients who are probably going to stay faithful once they switch, along these lines empowering the organizations to focus on their spending on clients who will create the most benefit.

Slide 14

Data Mining Applications: Other Applications Customer division All enterprises can exploit DM to find discrete fragments in their client bases by considering extra factors past conventional investigation. Fabricating Through decision sheets, makers are starting to redo items for clients; along these lines they should have the capacity to anticipate which elements ought to be packaged to take care of client demand. Guarantees Manufacturers need to foresee the quantity of clients who will submit guarantee claims and the normal cost of those cases. Visit flier motivators Airlines can recognize gatherings of clients that can be offered impetuses to fly more.

Slide 15

Data Mining in CRM: Customer Life Cycle Customer Life Cycle The phases in the relationship between a client and a business Key stages in the client lifecycle Prospects: individuals who are not yet clients but rather are in the objective market Responders: prospects who demonstrate an enthusiasm for an item or administration Active Customers: individuals who are as of now utilizing the item or administration Former Customers: might be "awful" clients who did not pay their bills or who brought about high costs It\'s critical to know life cycle occasions (e.g. retirement)

Slide 16

Data Mining in CRM: Customer Life Cycle What advertisers need: Increasing client income and client gainfulness Up-offer Cross-offer Keeping the clients for a more extended timeframe Solution: Applying information mining

Slide 17

Data Mining in CRM DM decides the conduct encompassing a specific lifecycle occasion Find other individuals in comparable life arranges and figure out which clients are taking after comparative conduct designs

Slide 18

Data Mining in CRM (cont.) Data Warehouse Customer Profile Data Mining Customer Life Cycle Info. Crusade Management

Slide 19

Data Mining in CRM: More Building Data Mining Applications for CRM by Alex Berson, Stephen Smith, Kurt Thearling (McGraw Hill, 2000).

Slide 20

Data Mining Techniques

Slide 21

Two Good Algorithm Books Intelligent Data Analysis: An Introduction by Berthold and Hand The Elements of Statistical Learning: Data Mining, Inference, and Prediction by Hastie, Tibshirani, and Friedman

Slide 22

Honest Tridas Vickie Mike Crooked Barney Wally Waldo Predictive Data Mining

Slide 23

Vickie Mike Tridas Prediction Honest = has round eyes and a grin

Slide 24

Decision Trees Data height hair eyes class short blond blue A tall blond brown B tall red blue A short dark blue B tall dark blue B tall blond blue A tall dark brown B short blond brown B

Slide 25

{tall, blue = A } short, blue = A tall, cocoa = B tall, blue = A short, chestnut = B short, blue = B tall, blue = B tall, brown= B Decision Trees (cont.) hair dim fair red Does not totally order blonde-haired individuals. More work is required Completely characterizes dim haired and red-haired individuals

Slide 26

{tall, blue = A } short, blue = A tall, chestnut = B tall, blue = A short, cocoa = B short, blue = B tall, blue = B tall, brown= B short = A tall = A Decision Trees (cont.) hair dull light red Decision tree is finished on the grounds that 1. Each of the 8 cases show up at hubs 2. At every hub, all cases are in a similar class (An or B) eye blue chestnut tall = B short = B

Slide 27

hair dull light red B An eyes blue cocoa A B Decision Trees: Learned Predictive Rules

Slide 28

Decision Trees: Another Example

Slide 29

Rule Induction Try to discover standards of the shape IF <left-hand-side> THEN <right-hand-side> This is the turn around of an administer based specialist, where the guidelines are given and the operator must act. Here the activities are given and we need to find the standards! Commonness = likelihood that LHS and RHS happen together (some of the time called "bolster calculate," "influence" or "lift") Predictability = likelihood of RHS given LHS (once in a while called "certainty" or "quality")

Slide 30

Association Rules from Market Basket Analysis <Dairy-Milk-Refrigerated>  <Soft Drinks Carbonated> pervasiveness = 4.99%, consistency = 22.89% <Dry Dinners - Pasta>  <Soup-Canned> predominance = 0.94%, consistency = 28.14% <Dry Dinners - Pasta>  <Cereal - Ready to Eat> pervasiveness = 1.36%, consistency = 41.02% <Cheese Slices >  <Cereal - Ready to Eat> pervasiveness = 1.16%, consistency = 38.01%

Slide 31

Use of Rule Associations Coupons, rebates Don\'t give rebates on 2 things that are habitually purchased together. Utilize the markdown on 1 to "force" the other Product situation Offer related items to the client in the meantime. Expands deals Timing of cross-showcasing Send camcorder offer to VCR buyers 2-3 months after VCR buy Discovery of examples People who purchased X, Y and Z (yet no combine) purchased W over a fraction of the time

Slide 32

Finding Rule Associations Algorithm Example: shopping for food For every thing, number # of events (say out of 100,000) apples 1891, caviar 3, dessert 1088, … Drop the ones that are underneath a base bolster level apples 1891, frozen yogurt 1088, pet nourishment 2451, … Make a table of every thing against each other thing: Discard cells beneath bolster limit. Presently make a 3D square for triples, and so forth. Include 1 measurement for every item on LHS.

Slide 33

Clustering The specialty of discovering gatherings in information Objective: accumulate things from a database into sets as per (obscure) basic qualities Much more troublesome than order since the classes are

View more...