Big Data & Artificial Intelligence ----How to Achieve Accurate Sales

Transcription

1 Big Data & Artificial Intelligence ----How to Achieve Accurate Sales Prof. Guangxia Xu Chongqing University of Posts and Telecommunications, Chongqing, China 1/30

2 Outline 1. Background 2. How to Achieve Accurate Sales 3. Applications in Other Industries 4. Future outlook 2/30

3 Outline 1. Background 2. How to Achieve Accurate Sales 3. Applications in Other Industries 4. Future outlook 3/30

4 1.Background Sales to brand strategy Internet technology and sales integration AI & Big Data and sales Fusion 4/30

5 1.Background Update Sales Strategy In Time Significance and Function Higher Performance Price Ratio Reduce Enterprise Sales Cost Enhancing Customer Satisfaction 5/30

6 1.Background Data Processing Takes A Long Time Error In Analysis Result Data Filtering Is Difficult Data Processing Error Data Quality Is Too Low The Plight of Accurate Sales 6 Data Value Is Not Represented Miss the Best Time of Sales 6/30

7 1.Background Bank Insurance Online Shopping 显示 AI 7/30

8 1.Background Data Collection AI Analysis Result Accurate Sales Data Preprocessing 8/30

9 9/30 1.Background Marketing Closed Loop Process 06 Marketing he Evaluation Effect 05 User Match 01 User Data Filtering 04 Match the Product 02 User Data Analysis 03 User Feature Extraction

10 Outline 1. Background 2. How to Achieve Accurate Sales 3. Applications in Other Industries 4.Future outlook 10/30

11 2. How to Achieve Accurate Sales Accurate Sales Platform Industry Customers Insurance Industry Financial Industry Fourism Industry Other Industry Sales Demand Recommendation System Accurate Sales Industry Users Data Cleaning Mail Recommendation User Recommendation Insurance Industry SMS Recommendation User portrait extraction technology Financial Industry User Management Social Software Recommendation Fourism Industry pre-trained deep neural network Ability Output Ability Output Sales Users Group Own Channel Sales Business Market Report Big Data Platform 11/30

12 2. How to Achieve Accurate Sales Application Layer Marketing Potential User Value Mining Loss of Customer Retention Value Customer Conversion Personalized Recommendation Stock User Interaction Combined Marketing Model building Feature Variable Selection Correlation Analysis Identify Smart Variables Building A Mining Model Output Target Group Business Layer User Portrait Basic Attributes Financial Credit Hobby APP Preference Location Preference Data Layer Data Collection Information Exchange Terminal Information Financial Credit APP Application Electricity Consumption 12/30

13 2. How to Achieve Accurate Sales Step1: Data cleansing 13/30

14 Data cleansing 14/30

15 Data cleansing Graph Data Processing V1 Storage Target Data Set V4 V2 V3 V2 V3 V5 V4 V5 V5 V4 Graph Data 15/30

16 2. How to Achieve Accurate Sales Step2: Personas 16/30

17 Personas Crawling Step1:Data Preprocessing Cleaning and Filtering User Demographics Text Data Network Data Features One Features Two Personas Features Three Features... Step2:Features Extraction 17/30

18 Personas Personas Personas A High-level Features Personas B Low-level Features Multi-source data Personas C Personas D 18/30

19 Personas PMI Characteristic Distribution Latent Dirichlet Allocation Keyword Vector User Feature Database Data Processing Data Cleansing User Initial Data 19/30

20 Personas User Feature Database Feature Sequence Feature Selection Feature Sampling User Feature Group Residual CHaracteristics PMI User Portrait 20/30

21 2. How to Achieve Accurate Sales Step3: Recommendation model 21/30

22 Recommendation model {W1,b1} {W2,b2} Input x h1 h2 Output y Hidden layers 22/30

23 Recommendation model Unsupervised Pre-training Autoencoder 1... Autoencoder 2... Autoencoder Supervised Fine-tuning /30

24 Recommendation model Personas X1,X2 XN(random) Y1,Y2 YN g^ DNN (w,b) 24/30

25 Recommendation model SAE Wage earner Stratum 25/30

26 Outline 1. Background 2. How to Achieve Accurate Sales 3. Applications in Other Industries 4.Future outlook 26/30

27 3.Applications in Other Industries Medical Industry Target Users Big Data + Artificial Intelligence Intelligent Diagnosis Drug Selection Equipment Purchase Personnel Employed Doctors Distribution 27/30

28 3.Applications in Other Industries Financial Industry Big Data + Artificial Intelligence Target Users Equity Market Insurance Industry VC Industry 28/30

29 Outline 1. Background 2. How to Achieve Accurate Sales 3. Applications in Other Industries 4.Future outlook 29/30

30 4.Future outlook Statistics Data Mining Database Machine Learning 1. In the Personas construction, the method of Machine Learning is introduced to adjust data parameters. 2. Preprocessing the data to avoid the curse of dimensionality. 3. Integrated use of cross domain data to break data dependencies. How to solve? Application Single Sample Deviation Timeliness Useless Data Inefficient 30/30

31 Thank You Q & A