Predictive Analytics Solutions, Edsson 2019
- 2. Predictive analytics based on large amounts of data Forecasting the future behavior of objects and subjects of analytical data to making optimal decisions.
- 3. Objectives of predictive analytics use: Fraud Detection Combining several methods of analysis makes it possible to improve the detection of models and prevent criminal behavior. As cybersecurity becomes an increasingly severe problem, high-performance behavioral analytics treats all online activities in real time to identify abnormalities that may indicate fraud. Marketing Optimization Predictive analytics is used to determine the answers of customers or purchases, as well as to promote cross-selling opportunities. Prognostic models help companies attract, retain and develop their most profitable customers. Operations Improvement Many companies use predictive models for forecasting stocks and managing resources. Airlines use predictive analytics to determine the price of tickets. Hotels are trying to predict the number of guests for any period to maximize attendance and increase revenue. Predictive analytics allows organizations to function more effectively. Risk Reduction Credit points are used to estimate the buyer's default probability for purchases and are a well-known example of predictive analytics. A credit score is a number generated by a predictive model that includes all data relating to the credit worthiness of a person. Other risk-related kinds of use including insurance claims and collections.
- 4. Areas of use Business and Marketing ● Segmentation of the market ● Forecasting demand, sales ● Selection of the best suppliers ● Comparative analysis and prediction of commodity prices ● Estimation of advertising effectiveness
- 5. Areas of use Finance and Banking ● Risk assessment ● Credit scoring ● Forecasting the outflow of the customers ● Fraud identification ● Forecast of account balances ● Monitoring of financial indicators
- 6. Areas of use Telecom / Internet ● Predicting network loads ● Classification of users ● Estimating the power of the necessary equipment ● Distribution of traffic ● Analysis of the content of Internet resources ● Planning of promo actions
- 7. Areas of use Medicine and Pharmacology ● Survival analysis ● Evaluation of the effectiveness of drugs ● Planning for medical research ● Analysis of the results of clinical research ● Disease Risk Assessment
- 8. Areas of use Production ● Quality control ● Process Monitoring ● Reliability analysis ● Planning of industrial experiments ● Analysis of the causes of quality loss ● Ensuring process stability
- 9. 1. Current sales level 2. Number of customers through channels 3. Number of debtors / departed clients 4. Number of defective products 5. Expenses for marketing campaigns 6. The quantity of products in order 1. Sales forecast for the week / month / year 2. Customer profiles on behavior / response 3. Predicting fraud / creditworthiness 4. Preventing process disruption 5. Forecasting the effect of marketing campaigns 6. Analysis of the basket, tips for cross-selling Descriptive Analytics Predictive Analytics What the difference?
- 10. What tools use? Descriptive Analytics Predictive Analytics ● Reporting ● OLAP ● KPI ● Dashboards ● Applied Statistics ● Applied Statistics ● Data mining ● Machine learning ● Modeling
- 11. What questions answer? Descriptive Analytics Predictive Analytics ● What have happened? ● Why did it happen? ● What will happen? ● When will happen?
- 12. The methods used to create predictive analytics
- 13. Used regression analysis methods: The core of the technique is a regression analysis that predicts the corresponding values of several correlated variables and is based on the proof or refutation of a specific assumption.
- 14. Used regression analysis methods: The core of the technique is a regression analysis that predicts the corresponding values of several correlated variables and is based on the proof or refutation of a specific assumption.
- 15. Used regression analysis methods: The core of the technique is a regression analysis that predicts the corresponding values of several correlated variables and is based on the proof or refutation of a specific assumption.
- 16. Regression Analysis Objectives 1. Determination of the degree of determinism of variation of a criterion (dependent) variable by predictors (independent variables) 2. Predicting the value of a dependent variable using an independent one. 3. Determination of the contribution of individual independentvariables to the variation of the dependent one.
- 17. Used methods of machine learning: The purpose of machine learning methods is not a direct solution to the problem, but training in the process of applying solutions to a set of similar tasks.
- 18. Used methods of machine learning: The purpose of machine learning methods is not a direct solution to the problem, but training in the process of applying solutions to a set of similar tasks.
- 19. Used methods of machine learning: The purpose of machine learning methods is not a direct solution to the problem, but training in the process of applying solutions to a set of similar tasks.
- 20. A non-linear model of dynamics of aggregated market prices of supply and demand Predictive Analytics of Purchasing Power
- 21. Predictive Analytics of Purchasing Power
- 22. Scheme of exchange and data processing
- 23. Stage №1: Analysis ● Resource management ● Suppliers ● Clients / customers ● Planning ● Quality control ● Monitoring ● Testing ● Marketing / Events ● Logistics ● Client base ● Support service ● Requests ● IT-Infrastructure ● Equipment park ● Machinery park ● Security Service ● Accounting ● Sales information ● Industry information ● Consumer reactions 1. Data collection
- 24. Stage №1: Analysis 1.1. Crawling Robot Robot for automatic data collection from various sources ● File system ● Branch sites ● Twitter, Facebook ● WebSites of competitors
- 25. Stage №1: Analysis 2. Extracting. Detection (parsing) ● People ● Places ● Organizations ● Dates References to other documents ● Biological termsa collection from various sources
- 26. Stage №2: Creating an Analytical Model The definition of the economic model, which is a functional dependence of the results of activity to the costs. Stage №3: Training Machine learning system for revealing statistical patterns based on regression analysis methods.
- 27. Result Forecasting the future behavior of objects and entities in order to make optimal decisions (for example, the price or demand for a product)