- Количество слайдов: 38
Summer 2014 Business Intelligence, Data Mining and Data Analytics/Predictive Analytics By: Asela Thomason
What is Business Intelligence • Basic Definition : Information that people use to support their decision making efforts. Data Mining and Data Analytics/predictive Analytics falls within this field.
What is Data Mining (The analysis step of Knowledge Discovery in Databases” Process or KDD), an interdisciplinary subfield of computer Science, is the computational process of discovering patterns in large data sets involving methods at the intersection of artificial intelligence, machine learning, statistics, and database management systems.
Basic-Definitions of Data Mining • The discovery of new, non-obvious, valuable information from a large collection of raw data • Data Mining (DM) is the core of the KDD [Knowledge Discovery in Databases] process, involving the inferring of algorithms that explore the data, develop the model and discover previously unknown patterns. • The set of activities used to find new, hidden or unexpected patterns in data
Definitions of Data Mining The detection of patterns from existing data. pattern n. (păt’ ərn) 1. A consistent, trait, feature, or method. 2. Any combination of values that contain meaning within the context or domain for which they are being reviewed
Data Mining -continued The overall goal of the data mining process is to extract information from a data set and transform it into an understandable structure for further use: (Predictive analytics) Discovering meaningful new corrections, patterns, trends. Example : Forecasting
Data Analytics/Predictive Analytics Data analytics (DA) is the science of examining raw data with the purpose of drawing conclusions about that information. Data analytics is used in many industries to allow companies and organization to make better business decisions and in the sciences to verify or disprove existing models or theories
Data analytics is distinguished from Data mining by the scope, purpose and focus of the analysis. Data miners sort through huge data sets using sophisticated software to identify undiscovered patterns and establish hidden relationships. Data analytics focuses on inference, the process of deriving a conclusion based solely on what is already known by the researcher.
Predictive analytics - Focus Uses lower level of Granularity, meaning it looks at the individual level. Instead of looking at which candidate will win the Presidential election in the state of Ohio, which is forecasting. It looks at the individual level. Which person is voting for or against. Predicts which individuals can be persuaded, which ones will not change, etc. Now with this information we ca change the outcome of the race. Obama used this technique very well.
Emerging Technology Data mining is one of the “ 10 emerging technologies that will change the world” listed by the MIT Technology Review (Larose). There is no doubt why many firms embrace data mining in their operations. An article in Information System Management points out that “data mining has become a widely accepted process for organizations to enhance their organizational performance and gain a competitive advantage”
Data Mining: Business • What is it? • Decision making • Marketing • Detecting Fraud • This technology is popular with many businesses because it allows them to learn more about their customers, prevent frauds and identity theft, and also make smart marketing decisions
Keys to a Successful Data Mining Project • Credible source of data • Knowledgeable personnel • Appropriate algorithms
Primary Tasks of Data Mining Classification classify a data item into one of several predefined classes Regression map a data item to a real-value prediction variable Clustering identify a finite set of categories or clusters to describe the data find a compact description for a set (or subset) of data Summarization Dependency Modeling describe significant dependencies between variables or between the values of a feature Change and Deviation Detection Discover the most significant changes
Some of the commonly used data mining methods are: • Statistical Data Analysis • Cluster Analysis • Decision Trees and Decision Rules • Association Rules • Artificial Neural Networks • Genetic Algorithms • Fuzzy Sets and Fuzzy Logic
Data Mining Applications In direct marketing a company saves much time by marketing to prospects that would have the highest reply rate. Instead of random selection on which customers to pick for their surveys, a company could use direct marketing from data mining to find the “correct” customers to ask.
Direct Marketing-Example 1 million mailers- cost $. 40 to ship letter=400, 000 cost Conversion is 1 percent without data mining
Direct Marketing using data mining, gives us 3% Conversion • Identifies smaller group, example ¼ of population and gets a higher conversion, 3% ,
Data Mining Applications Market segmentation is used in data mining in order to identify the common characteristics of customers who buy the products from one’s company. With market segmentation, you will be able to find behaviors that are common among your customers. As a company seeks customer’s trends, it helps them find necessities in order to help them improve their business.
Data Mining Applications Customer churn predicts which customers will have a change of heart towards your company and join another company (competitor). Although customer churns are negative to one’s business, it allows the corporation to seek out the problem they are facing and create solutions.
Customer Churn • Example: Magazine subscriber • Ideas to keep customer: • Discount, coupons, etc.
Data Mining Applications Market basket analysis- involves researching customer characteristics in respect to their purchase patterns Example: Ralphs Club Card Cereal and Milk
Market Basket • Beer and diapers • merchandising
Prediction based on Data mining/Predictive analysis • Examples of real life. • Target – can predict which customers will be pregnant • Hospitals can predict which payments may need to be admitted • Credit card – can predict which customers may miss their payment based upon where card is used. Example Bar-alcohol=missed payments
Class Identification • Mathematical taxonomy • Concept clustering
Data Mining Applications Class identification, which consists of mathematical taxonomy and concept clustering. Mathematical taxonomy focuses on what makes the members of a certain class similar, as opposed to differentiating one class from another. For example, Ralphs can classify its customers based on their income or past purchases
Data Mining Applications Concept clustering - determines clusters according to attribute similarity. Consider the pattern a purchase of toys for age group 3– 5 years, is followed by purchase of kid’s bicycle within 6 months about 90% of the time by high income customers, which was discovered by data mining. The Company can identify the prospective customers for kid’s bicycle based on toy purchase details and adjust the mail catalog accordingly.
Data mining Applications Deviation analysis, A deviation can be fraud or a change. In the past, such deviations were difficult to detect in time to take corrective action. Data mining tools help identify such deviations. For example, a higher than normal credit purchase on a credit card can be a fraud, or a genuine purchase by the customer. Once a deviation has been discovered as a fraud, the company takes steps to prevent such frauds and initiates corrective action
Making Better Decisions • Patterns and trends • What to produce? • Equal Success
Privacy • Sensitive information Data mining increase incentives to get more sensitive data Seeing into private future- Target Do we have the right Employers try to predict churn
Response Bias Issues • Types o o –Coverage or frame error –Sampling error –Nonresponsive error –Measurement error • Flawed data
Data Mining in Medical The most recent and most promising use of data mining has been the development of data mining tools for the medical sector. The use of data mining to extract patterns from medical data provides near endless opportunities for symptom trend detection, earlier detection of illness, DNA trend analysis and improved patient reactions to medicines. These many advantages allow doctors and hospitals to be more effective and more efficient.
Advantages of Data Mining: Medicine • Earlier detection of illness • Symptom trends • Data analysis • Improved drug reactions
Disadvantages of Data Mining: Medicine • No uniform language - Medical • Incomplete records • Privacy
Data mining - Medical How data mining is actually used to analyze individual data can become quite complex due to the data. The goal of the process is to take the medical data which contain many attributes and determine which ones are actually relevant to the diagnosis, symptom or result. Two methods used in medical data mining are clustering, discussed previously and biclustering.
Bi. Clustering A new form of clustering called biclustering is now being used to help associate certain genetic patterns with illnesses. Biclustering for genetic research is that it does not simply assign a sample to a certain classification across the board it takes into account other variables which increases accuracy since not all genetic traits are active all the time, special conditions are sometimes necessary for traits to surface.
Study shows automated data mining surveillance helps detect blood culture contamination. (Clinical report) World Disease Weekly, Jul 11, 2006. Reading Level (Lexile): 1440 Med. Mined, Inc. , a medical information technology company, announced results of a study that showed how automated data mining surveillance helped in detecting and reducing an endemic issue of blood culture contamination in a hospital's hematology/oncology unit. The study was conducted at Hendrick Medical Center, a 453 -bed hospital in Abilene, Texas, that began using Med. Mined's Data Mining Surveillance service in June 2005. After incorporating data mining surveillance house-wide, the hospital discovered that, in patients who had a hospital stay of 3 or more days, blood isolates caused primarily from skin flora occurred with the third most frequency. Further, the unit with the highest rate of blood isolates was hematology/oncology. After implementing process improvements, results from October to December 2005 demonstrated a 50% decrease in the overall number of isolates from patients on the unit, with a 61. 5% decrease in isolates of skin flora. Med. Mined is a Birmingham, Alabama-based company founded in 2000 to provide data mining analysis and related technical, clinical, and financial consulting services to the healthcare community. Med. Mined's patented infection-tracking technology and proprietary service are being used by nearly 200 hospitals. This article was prepared by World Disease Weekly editors from staff and other reports. Copyright 2006, World Disease Weekly via News. Rx. com.
Net. Flix • Case and discussion
Using Internet to market a new Winery • Case • discussion