Download presentation
Presentation is loading. Please wait.
Published bySuzan Miller Modified over 8 years ago
1
Advancements in Analytics with Azure Machine Learning James Wang Technical Evangelist Microsoft Taiwan Slide modified from https://github.com/Azure-Readiness/hol-azure-machine-learning
2
Agenda What is Machine Learning What is Azure Machine Leaning (Hands-on) Azure Machine Learning Studio
3
What is Machine Learning ? https://pythonprogramming.net/machine-learning-python-sklearn-intro/
4
What is Machine Learning ? Using known data, develop a model to predict unknown data.
5
What is Machine Learning ? Using known data, develop a model to predict unknown data. Known Data: Big enough archive, previous observations, past data Unknown Data: Missing, Unseen, not existing, future data Model: Known data + ML Algorithms
6
From Learning to Machine Learning Human learning Machine learning observationlearningskilldataMLskill
7
From Learning to Machine Learning Machine learning What is skill for machine? Improving performance measure e.g. prediction accuracy dataMLskill
8
Skill
9
Microsoft & Machine Learning Bing maps launches What’s the best way to home? Kinect launches What does that motion “mean”? Azure Machine Learning GA What will happen next? Hotmail launches Which email is junk? Bing search launches Which searches are most relevant? Skype Translator launches What is that person saying? 201420091997201520102008
11
EXAMPLE
12
Classify a news article as (politics, sports, technology, health, …) Politics SportsTechHealth Model (Classification) Using known data, develop a model to predict unknown data.
13
Known data (Training data) Using known data, develop a model to predict unknown data. DocumentsLabels Tech Health Politics Sports Documents consist of unstructured text. Machine learning typically assumes a more structured format of examples Process the raw data
14
Known data (Training data) Using known data, develop a model to predict unknown data. Labels Documents Feature DocumentsLabels Tech Health Politics Sports Process each data instance to represent it as a feature vector Label Features Feature Vector
15
Feature vector Known data Data instance i.e. TF-IDF of key-words {0.40, 0.18, 0.08, 0.11, 0.7, 0.7, …..} : Health
16
Developing a Model Using known data, develop a model to predict unknown data. DocumentsLabels Tech Health Politics Sports Training data Train the Model Feature Vectors Base Model Adjust Parameters
17
Machine Learning Algorithms Supervised learning ( 監督式學習 ) Data with label This customer will like coffee This network traffic indicates a denial of service attack Unsupervised learning ( 非監督式學習 ) Data without label These customers are similar This network traffic is unusual
18
Common Classes of Algorithms Classification 分類 Regression 迴歸 Clustering 分群
19
What kind of algorithm you need? If you want to answer a YES|NO question, it is binary-classification If you want to answer a label question, it is multi-class-classification If you want to predict a numerical value, it is regression If you want to group data into similar observations, it is clustering
20
Classification 分類 Scenarios: Which customer are more likely to buy, stay, Which transactions|actions are fraudulent Which quotes are more likely to become orders Recognition of patterns: speech, speaker, image, movement, etc. Algorithms: Boosted Decision Tree, Decision Forest, Decision Jungle, Logistic Regression, SVM, ANN, etc. Classification
21
Clustering 分群 Scenarios: 所有分群問題 Algorithms: K-means Clustering
22
Regression 迴歸 Scenarios: Stock prices prediction Sales forecasts Premiums on insurance based on different factors Quality control: number of complaints over time based on product specs, utilization, etc. Workforce prediction Workload prediction Algorithms: Bayesian Linear, Linear Regression, Ordinal Regression, ANN, Boosted Decision Tree, Decision Forest Regression
23
What kind of algorithm you need? If you want to answer a YES|NO question, it is binary-classification ( 分類 ) If you want to answer a label question, it is multi-class-classification ( 分類 ) If you want to predict a numerical value, it is regression ( 迴歸分析 ) If you want to group data into similar observations, it is clustering ( 分群 )
24
Binary versus Multiclass Classification Does your customer want a yes|no answer? Binary examples click prediction yes|no over|under win|loss Multiclass examples kind of tree kind of network attack type of heart disease
25
Regression versus Classification Does your customer want to predict|estimate a number (regression) or apply a label|categorize (classification)? Regression problems Estimate household power consumption Estimate customer’s income Classification problems Power station will|will not meet demand Customer will respond to advertising
26
Which algorithm performs better?
27
Model’s Performance Known data with true labels Tech Health Politics Sports Tech Health Politics Sports Tech Health Politics Sports Training data 80% Testing data 20% Model’s Performance Difference between “True Labels” and “Predicted Labels” True labels Tech Health Politics Sports Predicted labels Train the Model Split Detach Test trained model with features Compare prediction with true labels +/-
28
Azure Machine Learning
29
Microsoft Azure Machine Learning “Make machine learning accessible to every enterprise, data scientist, developer, information worker, consumer, and device anywhere in the world.” Data Science is far too complex, need a simpler/scalable method. Problem definition Data storage Feature selection Model evaluation Deploy to application
30
Blobs and Tables Hadoop (HDInsight) Relational DB (Azure SQL DB) Data Clients Model is now a web service that is callable Monetize the API through our marketplace API Integrated development environment for Machine Learning ML STUDIO
31
Azure Machine Learning Studio https://studio.azureml.net/
32
Load a Data Set Add Transformations and Filters Create the Experiment Path and apply Algorithms Save and Run the Model Publish the model Use the model Microsoft Azure Machine Learning
33
https://studio.azureml.net/
34
Thanks for your attention
Similar presentations
© 2025 SlidePlayer.com. Inc.
All rights reserved.