Transcript
DATA SCIENCEAFFECTLY & FITNESS
INTRODUCTION
1. Data Processing2. Machine Learning3. Model Evaluation4. Visualization & Reporting
• Getting Data• Exploratory Data Analysis• Cleaning Data• Transforming Data~ BI
2. MACHINE LEARNINGChoosing the right:• Problem• Model• Algorithm
• Predictive Analysis• Natural Language Processing• Segmentation• Recommendation...
2. MACHINE LEARNING (MODELS & ALGORITHMS)
• Boosting• AdaBoost• XGBoost• Gradient Boosting...
• Deep Learning• Artificial NN• Convulnational NN• Recurrent NN...
3. MODEL EVALUATION
k-fold cross-validation
AFFECTLY (SEGMENTATION)Problems:• Data • Resource (memory)• Reporting
AFFECTLY (SEGMENTATION)Problems:• Data • Resource (memory)• Reporting
AFFECTLY (PREDICTIVE ANALYSIS)
Problems:• Right model• Data
Top secret
FITNESS (PREDICTIVE ANALYSIS)
pretty simple
FITNESS (RECOMMENDATION)• Popularity – Based• Content – Based• Demographic – Based
(Facebook)• Collaborative Filtering• Hybrid systems
• Problems:• User_data• Item_data
• Approach:• User – User• Item – Item
• Algorithm:• k-NN• Latent Factors (MF)
• Feedback:• Explicit (k-NN)• Implicit (MF)
• Domain:• Single• Multiple
Q&A
Thank you for listening
top related