CompTIA DataX DY0-001 TTT Session 7:
September 04, 2024
@TeachCompTIA
1
Certificate of
Slides Bios Q&A Attendance Call to Action
Multimedia Today’s Resources ON24 Help Group Chat Survey
Copyright (c) 2024 CompTIA Properties, LLC. All Rights Reserved. | [Link] 2
DataX Team
Instructor: Host:
RJ Sak Stephen Schneiter
raymondjsak@[Link] Instructor Network Program Director
CompTIA
sschneiter@[Link]
Copyright (c) 2024 CompTIA Properties, LLC. All Rights Reserved. | [Link]
The CompTIA Instructor Network (CIN) is a worldwide community for instructors who
provide CompTIA certification training.
Benefits of being a community member include:
∙ Communicate and collaborate with CompTIA staff and other instructors.
∙ Access resources for students to understand the value of getting certified.
∙ Receive complimentary training and tools from CompTIA to enrich your classroom.
∙ Become proficient at teaching CompTIA standards.
∙ Share best practices and resources with each other.
[Link]
Complimentary Webinar Series for Instructors
What: 10-session webinar series
The CompTIA Tech+ FC0-U71 TTT series will cover:
When: Sept. 10 – Oct 09, 2024
• Tech+ exam domains Where: ON24
• Comprehensive understanding of key fundamental
technology concepts
• Hands-on experience with key technology tools used by
IT professionals
• Instructional strategy to implement a Tech+ course
• Preparation for the Tech+ FC0-U71 certification
Copyright (c) 2024 CompTIA, Inc. and its Affiliates. All Rights Reserved. | [Link]
DataX DY0-001 TTT Session Outline
Date Topic
Illustrating the Data Science Lifecycle
08/12/2024
Analyzing Business Problems
Collecting Data
08/14/2024 Cleaning and Preparing Data
08/19/2024 Describing Data Features
Exploring Data
08/21/2024
Navigating the Model Selection Process
08/26/2024 Employing Machine Learning Methods
08/28/2024 Experimenting With Deep Learning
09/04/2024 Evaluating and Refining Data Models
09/09/2024 Communicating for Business Impact
09/11/2024 Deploying Data Models
09/16/2024 Discovering Specialized Data Science Applications
Copyright (c) 2024 CompTIA Properties, LLC. All Rights Reserved. | [Link] 6
Optimization Techniques
Optimize Models and Resources
opyright © 2024 The Computing Technology Industry Association, Inc. All
C 7
rights reserved.
Optimization: How do algorithms calculate
models (formulas) for the data?
Sales
opyright © 2024 The Computing Technology Industry Association, Inc. All rights
C 8
reserved.
Temperature
Optimization: OLS Regression minimizes the
total distance across observations to the
trendline
Sales
opyright © 2024 The Computing Technology Industry Association, Inc. All rights
C 9
reserved.
Temperature
Gradient-Based Methods
Variants of Second-order
Gradient descent
gradient descent methods
Fundamental Momentum
Newton’s
optimization
method
technique AdaGrad
Uses the RMSProp
Hessian-based
gradient of the
methods
loss function Adam
opyright © 2024 The Computing Technology Industry Association, Inc. All rights
C 10
reserved.
Overfitting
Regularization Methods
L1 Regularization L2 Regularization Benefits
Adds absolute sum Adds squared sum
Penalize model
of parameters to of parameters to
complexity
loss loss
Encourages small, Prevent overfitting
Encourages
evenly distributed by discouraging
sparsity
weights complex models
opyright © 2024 The Computing Technology Industry Association, Inc. All rights
C 12
reserved.
Loss Functions
Multi-objective
Common loss functions
optimization
Mean squared error (MSE)
Pareto optimality
Cross-entropy loss
Weighted-sum methods
Hinge loss
opyright © 2024 The Computing Technology Industry Association, Inc. All rights
C 13
reserved.
Applying Optimization
Techniques in Scheduling and
Pricing
Optimize Models and Resources
opyright © 2024 The Computing Technology Industry Association, Inc. All
C 14
rights reserved.
Optimization in Scheduling
Define objectives and constraints
• Minimize completion time, maximizing resource usage
• Task dependencies, resource availability
Optimization techniques
• Linear Programming: used to find the best possible schedule
• Heuristic Algorithms: genetic algorithms for optimal scheduling
Implementation and benefits
• Specialized software or custom algorithms
• Improved resource usage, reduced costs, enhanced quality
opyright © 2024 The Computing Technology Industry Association, Inc. All rights
C 15
reserved.
Optimization in Pricing
Defining goals and data
• Goals: Maximizing revenue, profit margin, or market share
• Data: Historical sales figures, customer behavior, competitor pricing,
production costs
Building pricing models
• Demand Elasticity: How price changes affect demand
• Customer Segmentation: Identifying groups with different price
sensitivities
Optimization algorithms and testing
• Algorithms: Linear programming, machine learning techniques
• Testing: A/B testing to assess effectiveness and refine models
opyright © 2024 The Computing Technology Industry Association, Inc. All rights
C 16
reserved.
Resource Allocation and Bundling
Strategies Using Optimization
Techniques
Optimize Models and Resources
opyright © 2024 The Computing Technology Industry Association, Inc. All
C 17
rights reserved.
Cloud Computing Platforms
• Scalable resources
▪On-demand access to
CPUs, GPUs, TPUs
• Preconfigured
environments
▪Optimized environments
for machine learning
Images © [Link]
Copyright © 2024 The Computing Technology Industry Association, Inc. All rights reserved. 18
Containerization and Orchestration
• Docker containers
▪Packages models and dependencies
▪Facilitates deployment
▪Provides consistent environments
• Kubernetes orchestration
▪Manages containerized deployments
▪Automates scaling
▪Optimizes usage
opyright © 2024 The Computing Technology Industry Association, Inc. All rights
C 19
reserved.
Monitoring and Optimization Tools
• Resource monitoring tools
▪Tracks resource use
▪Identifies bottlenecks
▪Monitors CPU, GPU and memory
• AutoML platforms
▪Automates ML pipeline
▪Automates resources allocation
▪Performance optimization
opyright © 2024 The Computing Technology Industry Association, Inc. All
C 20
rights reserved.
Lesson 10.2
Explain Optimization Problem Types
Lesson-Exam Objective Alignment
• 5.1 Compare and contrast optimization concepts.
opyright © 2024 The Computing Technology Industry Association, Inc. All
C 21
rights reserved.
Explain Optimization Problem Types
• Linear and non-linear solvers in optimization
• Handling boundary cases and unconstrained
optimization techniques
• Advanced topics in optimization: Bandit problems and
local maxima/minima
opyright © 2024 The Computing Technology Industry Association, Inc. All rights
C 22
reserved.
Linear and Nonlinear Solvers
Linear solvers
• Methods: Direct and iterative
• Strengths: Speed and efficiency
• Limitations: Computationally expensive for large systems
Non-linear solvers
• Methods: Gradient-based and derivative-free
• Strengths: Broader applicability
• Limitations: Computationally expensive; need careful selection
opyright © 2024 The Computing Technology Industry Association, Inc. All rights
C 23
reserved.
Handling Boundary Cases
Determine
Problem Implement Testing and
handling
Analysis custom logic validation
techniques
opyright © 2024 The Computing Technology Industry Association, Inc. All rights
C 24
reserved.
Use Cases for Handling Boundary Cases
Solving subproblems Designing complex
Exploring solution
in complex machine learning
space
optimization models
• Complex optimization • Understanding the • Understanding the
problems involving behavior of the behavior of the
multiple stages or objective function objective function
hierarchies
opyright © 2024 The Computing Technology Industry Association, Inc. All rights
C 25
reserved.
Bandit Problems
Exploration vs.
Algorithms Applications Limitations
Exploitation
Exploration: A/B testing
Gather Data collection
Epsilon-Greedy
information using challenges
various options
Recommendation
systems
Exploitation:
Algorithm
Choose best Upper Confidence
effectiveness
option based on Bound (UCB)
Online advertising varies
observations
opyright © 2024 The Computing Technology Industry Association, Inc. All rights
C 26
reserved.
Local Maxima and Minima
Machine Learning Business
Techniques
Applications Applications
• Grid search • Hyperparameter • Optimal asset
• Random search optimization allocation
• Gradient descent • Monitoring training • Supply chain
process optimization
opyright © 2024 The Computing Technology Industry Association, Inc. All rights
C 27
reserved.
Lesson 10.3
Tune Hyperparameters
Lesson-Exam Objective Alignment
• 2.4 Given a scenario, conduct a model design iteration process.
• 3.1 Given a scenario, apply foundational machine-learning conc
opyright © 2024 The Computing Technology Industry Association, Inc. All
C 28
rights reserved.
Tune Hyperparameters
Hyperparameters
- Set before training and control the learning process
- Examples: learning rates, number of hidden layers
Model parameters
- Learned from the data during training
- Example: weights in a neural network
opyright © 2024 The Computing Technology Industry Association, Inc. All rights
C 29
reserved.
Accuracy of Predictions for Different Models
Regression Models
R-squared Residual
Error metrics
analysis analysis
• Mean Squared • R-Squared • Visualizing
Error (MSE) • Adjusted R- Residuals
• Root Mean Squared • Goodness-of-Fit
Squared Error tests
(RMSE)
• Error (MAE)
differences
opyright © 2024 The Computing Technology Industry Association, Inc. All rights
C 30
reserved.
Accuracy of Predictions for Different Models
Classification Models
Overfitting
Error Metrics Visualization
Mitigation
• Accuracy • Pruning • Feature
• Confusion • Cross-validation importance
matrix analysis
opyright © 2024 The Computing Technology Industry Association, Inc. All rights
C 31
reserved.
Activity – Building a Regression Model
• Determine the methods you
would use to build a linear
regression model to predict
house prices.
• Explain the steps you would
take and how you would
evaluate the data.
Images © [Link]
Copyright © 2024 The Computing Technology Industry Association, Inc. All rights reserved. 32
Activity – Building a Regression Model
Steps:
1. Split the dataset into training and testing sets
2. Train the linear regression model on the training set
3. Predict housing prices on the testing set
Evaluation
4. Calculate MSE, RMSE, and MAE
5. Analyze R-Squared and Adjusted R-Squared
6. Visualize residuals and perform goodness-of-fit tests
Images © [Link]
Copyright © 2024 The Computing Technology Industry Association, Inc. All rights reserved. 33
Summary
Evaluating and Refining Data Models
opyright © 2024 The Computing Technology Industry Association, Inc. All
C 34
rights reserved.
Main Points
• Optimization techniques improve model effectiveness and
resource use, benefiting allocation, scheduling, pricing, and
network design, with benchmarking as foundational support.
• Analyze results to justify model recommendations and selections,
ensuring effective and efficient solutions using appropriate
optimization methods.
• Compare and contrast optimization concepts to select appropriate
methods and tools for effective solutions, enabling the resolution
of complex decision-making problems across various fields.
opyright © 2024 The Computing Technology Industry Association, Inc. All
C 35
rights reserved.
Discussion time: Please type your questions in chat
Questions over content.
Share you experience.
What would you like to see
different moving forward?
Let’s keep the conversation going in the CompTIA Instructor
Forum: [Link]
Copyright (c) 2024 CompTIA Properties, LLC. All Rights Reserved. | [Link] 36
