Finishing Business Section in AI Course - Day 18 at Techie Youth
Wednesday - Hours Worked: 5 hrs and 30 mins
Today I finished the entire Business Section in the AI course by reading all of the remaining articles provided. It was a very interesting read and I learned alot about the potential use cases of different AI models for business. I then started the assignment by checking out some databases provided by the State of New York. Here is what I covered today:
Finished Big Data and IOT (sub - section) by reading the following articles:
* 4 Ways that Big Data can help businesses of any size identify big opportunities
1) Predictive Analytics - helps analyze trends to predict what is most likely going to happen in the next few weeks, months, or years. This can help businesses manage supplies and resources.
2) Customer Insights - predictive analytics can help retailers understand shopping habits and behaviors of customers. It can determine what customers are most likely to buy on their first visit, what they will buy on every other visit, and many more.
3) Marketing Opportunities - this can come from recognizing an increase in sales of a certain product and then tracing it to its source. By knowing who is buying these products, a business can better market their product towards that target audience giving them better marketing opportunities thanks to Big Data.
4) Effective Production Management - many businesses can struggle with managing many different projects at the same time. Using Big Data can help effectively manage all of the projects and which stages of completion they are in.
* There are many applications for Data Analysis including manufacturing, predictive maintenance processes, enterprise resource planning systems, supplier relationship management systems, and product life cycle management systems.
* Data Analysis can make all these processes and systems more efficient and help businesses grow.
* Big Data is now being used in many industries for manufacturing and has been used in many cases to improve certain aspects of a business from recycling waste to managing automation of tasks/ labor.
Finished Neural Analytics (sub - section) by reading the following articles:
* Artificial Neural Networks (ANNs) provide an advantage over normal computers in their ability to learn and adapt. But to do this they need an incredible amount of data
* ANNs can be used for a variety of tasks but it's currently most popular in pattern recognition which ranges from optical character recognition to facial recognition. This is great for many industries including agriculture, design, and the beauty industry which can greatly benefit from finding patterns in images.
* In the business field, ANNs have been used for marketing, consulting services, prediction, and detecting frauds.
1) Marketing Strategies: NNs can segment customers to different classifications based on characteristics and better market products that these customers would buy
2) Improving Search Engine Functionality: NNs have been used to help reduce search errors. Through enough training this accuracy can become even better.
3) Pharmaceutical Industry: Used for disease identification and diagnosis as well as creating better treatment plans
4) Retail Sector: NNs are used for their forecasting ability to help businesses manage their stock appropriately
5) Repeat Custom: NNs can help remind customers to buy an item to increase the frequency of them coming back to purchase said item. This can help keep customers loyal to your business.
6) Financial Services: NNs can help predict marketing movements, improve the way banks operate, and help identify credit risks
7) Insurance Provision: NNs allows for segmentation of policyholders which allows for better pricing plans
8) Fraud Detection: NNs can be used to identify unusual activity and prevent frauds.
9) Store Layout: NNs can better optimize store layout so customers can purchase more products from your store.
10) Facial Recognition: NNs can be used for business security through facial recognition. This can help detect shoplifters and criminals that enter your business.
Checked out “Covid Data”, “Education”, and “Economic Development” databases on data.ny.gov.
Learning How to Host A Website - Day 19 at Techie Youth
Friday - Hours Worked: 7 hours
Today I finished the first section (“Hosting your website) in the Website Administration section. It was very helpful in giving me an insight on how to upload a website. This information will definitely be useful for uploading my Portfolio in the future. This is what I finished today:
Finished Video: “Web Hosting Tutorial For Beginners” - Ray DelVecchio
* To access a website the user enters the domain of a website. This then gets passed through DNS Servers which converts a website into an IP address for computers to look for the host.
* You can use GoDaddy to register a domain. Then use HostGator to get a DNS server. When you get this server, go to your GoDaddy account and plug in the servers into DNS settings.
* Types of Web Hosting:
Finished Video: “Host Your Website For Free Tutorial” - Dev Ed
* Free hosting site (static website): Netlify
* Sign up for Netlify, drag a website inside a folder into the sites page and your website should be deployed with a random domain.
* You can also set up a custom domain but if you run into a privacy issue it could be from the site you go your domain from
* The video uses a site called “namecheap” and he adds an “A Record” and “CNAME RECORD”. The first one stores the IP address and the other is the address domain generated by Netlify. It may not work right away and you’ll have to wait for the servers to update.
* You can make updates to the site by dragging the folder with your changes onto Netlify and they will take care of the rest.
FInished Video: “How to Easily Upload a Website” - Dani Krossing
* Learned how to sign up for HostGator and learned about what I would be paying for.
* Always check over your website before uploading such as spelling and correct format depending on size.
* Use validator.w3.org to check if your website can be run on multiple browsers.
* Learned about FileZilla which basically helps us upload the files of your website.
Finished Video: “How to Put Your Website Online” - LearnCode.Academy
* Video was basically a review on what I learned in previous videos. Video mentions the use of hostgator.com for web hosting and the use of cyberduck to upload web files.
* Learned how to use cyberduck to transfer website files to the server. This requires the server IP, Username, and Password from hostgator. Copy files to www folder. Make sure to delete the default html file if there is one.
* Found this video to be more easy to follow so might use this when uploading a website
Finished Short Video: “What is a Domain Name?” - Create A Pro Website
* A domain name is just an address to your website. It’s a name for your IP address so people can look for it easier
Finished Video: “How a DNS Server (Domain Name System) Works.” - PowerCert Animated Videos
* A DNS resolves domain names to IP addresses. When you type in a website name, the DNS server will look for an IP address that matches that website name. After that your computer will then be able to connect to the web server that is hosting that website and retrieve information about that website.
* When looking for a website your computer will send a command to a Resolver which then communicates with other servers to find the website. It will first communicate with the “Root” server which will direct the resolver to a “Top Level Domain” (TLD) server and then to a “Authoritative Name” server that gets the IP address. The IP address is sent to the Resolver and then to your computer which can now access the certain web page.
Practiced some HTML by building the sample site I learned from scratch.
REUPLOAD
Monday - Hours Worked: 2 hrs
I decided to move back to the AI course today and covered the basics of how to start a Robotics company. It was quite interesting but I don’t think this is something I’m interested in. Sounds way too stressful! Here’s the only thing I did today:
Finished article on how to start a robotics company:
In modern times the push for automation has increased exponentially. This societal shift has been great for robotics businesses as they embrace this change to automated tasks. Here is a TLDR of 10 steps to starting a robotics company.
1 - Plan your business (Very vital part)
Setting a clear plan for creating a company is vital. It will help map out specifics of the business. Questions that you should ask yourself is:
* What are the costs of starting the business?
Launching a business requires a lot of upfront costs. These can include office space, computers, raw materials, and/ or employees. Make sure that you can cover these before starting a business.
* What will the ongoing costs for your company be?
To run your company you will need to take in account the expenses needed to run the company. Robotics businesses tend to use a lot of software that require subscriptions and materials that may need to be restocked every so often after every project. Make sure to keep in mind that you need to pay your employees! They can get expensive and need money to fund their research.
* The target market?
Most businesses tend to create a product to appeal to the general market. Think about how attractive your product will be to the market.
* How does your company make money?
Robotics companies make money by selling or renting products to other companies. They can also sell their products for consumer use as well.
* How much will you charge customers?
If you charge your customers too much they will be hesitant to buy. It’s important to consider if your product is similar to other ones in the market. If it is then it's probably better to charge the market price or (if your company can afford to) charge it lower than the market price for more competition. If your product is completely new, you should figure out a price that can be attractive to many consumers.
* What is the name of your business?
Names have a huge impact due in modern day so make sure that the name can be somewhat memorable.
2 - Form a legal entity
Establishing a legal business entity can protect you when your company is sued. You can form an LLC(Limited Liability Company) or hire an LLC service
3 - Register for Taxes
You need to register for state and federal taxes before you can open your business.
4 - Open a Business Bank Account & Credit Card
Creating these are vital for personal asset protection in the event you are sued.
5 - Set Up Business Accounting
Recording various expenses of your business and sources of income is vital to understand the financial position of your business. This can greatly simplify annual tax filing.
6 - Obtain Necessary Permits and Licenses
Make sure to meet State & Local Business Licensing Requirements and have a Services Contract, an Informed Consent Agreement, Labor Safety Requirements, and Certificate of Occupancy.
7 - Get Business Insurance
Just like how you need Insurances, a business needs one as well in order to operate safely and lawfully. It can protect your company’s financial wellbeing in the event of a covered loss. You can begin with the General Liability Insurance which is the most common coverage that small businesses need.
8 - Define Your Brand
A brand is what a company represents as well as how it is perceived by the public. A strong branding can help stand out from competitors. If your company is producing consumer goods, branding becomes vital for profit. Getting a good marketing team can help your company attract more customers and having improved products that come out in an establish time period can help keep customers.
9 - Create A Business Website
Create a site that describes your company's goals and ideals. Make it aesthetically attractive to capture a user's attention.
10 - Set Up Your Business Phone System
Setting up a phone system can help you separate your personal life from your business. It can help your business by making things more automated and easier for potential customers to find and contact you.
(From howtostartanllc.com)
Hours Worked: 6hr 30 mins
Today I set out to finish the “Getting Started Creating Websites” section. I have to say this section was a lot of fun and I enjoyed playing around with CSS while following along with the tutorial. Here's what I did for the day:
Finished tutorial video on how to build a layout:
* Learned about section tags and footer tags
* Learned the difference between margin and padding as well as how to use them in CSS
(From LearnCode.Academy)
Finished tutorial video on building CSS layout with Flexbox:
* Learned some VSCode shortcuts for HTML development such as doing div*3 to create 3 div tags. This works for other tags as well.
* Learned how to use display in CSS to orient tags in a specific manner (e.g inline, inline-block, flex)
* When using inline-block you can edit it’s dimensions by entering specific heights and widths
* Instead of using inline-block you could use flex in the CSS of the parent tag to display tags in a inline-block format
* Learned how to use flex-direction to change orientation of flex display
* By using margin auto, CSS will automatically add space between tags
* Learned how to create a top navigation using ul and li tags
* By setting list-style-type to none we can get rid of the bullet points that come default with li tags
(From LearnCode.Academy)
Finished tutorial video on HTML/CSS with multiple pages:
* Learned the properties of an a tag and created links to different pages
* If you have a website with many pages, be sure to create folders to store them. Organization is key!
* Learned how to create an external CSS file and link it to a HTML file. This way my HTML pages don’t get too long when coding since CSS can be multiple lines.
*
(From LearnCode.Academy)
Finished tutorial video on how to use Images when creating website:
* Learned how to use background-image in CSS to include any photo I want for the specific tag
* Learned how to use background-size to edit the size of the image and background-repeat to determine if the image repeats (happens when we decrease the size of the image too much and HTML tries to repeat the image to fill in spaces in the tag)
* Learned how to use position: relative
* Learned how to use img tag in HTML to insert images
* Learned how to use border-radius in CSS to round images
* Learned how to use box-shadow in CSS to add depth to images
* Learned how to use figure-caption tag for images in HTML
(From LearnCode.Academy)
Finished tutorial video on how to create better responsive websites:
* Learned to use @media screen and (max-width: ## ) to determine what my website would look like, depending on the screen size. This ensures that my website looks good on any device.
* Use @media screen and (max-width: ## ) for desktop first development and @media screen and (min-width: ## ) for mobile first development
* With each @media screen we can add new blocks of CSS for a different design based on the screen size.
* Learned to use display: block in CSS
(From LearnCode.Academy)
Finished Quiz for this section.
After finishing these tutorial videos I made sure to practice everything I learned by creating a similar formatted website with slight differences. I created a slide navigation bar instead of it being on top and used different images than the one provided by the video. I also used different fonts for the text as well as different colors. Although I feel like I have a good grasp of information learned, I think practicing a bit more on my next work day will be great for fully mastering the material I covered today.
Friday - Time Worked: 5 hours and 30 mins
Today I decided to bounce back into the AI course and decided to hop on to the Creating a Business Section. It was not as interesting as I thought it would be unfortunately. I guess I’m not really interested in any business side of things when it comes to AI, but I was glad to learn about it nonetheless. So here’s what I covered today:
Finished reading “120 Machine learning Business Ideas” Blog
* Machine learning is on the edge of revolutionizing many fields in the world right now. In the short term, Machine Learning can have varying levels of impact on different fields. This includes:
Automotive, Manufacturing, Retail, Finance, Agriculture, Energy, Health-care, Pharmaceuticals, Public & Social, Media, Telecom, and Logistics.
I found that the data provided in this blog to be very interesting. For example, I found that Machine Learning can have a great impact on predictive maintenance (1.6) with a moderate amount of data to use (0.7). I found it surprising that Machine Learning wouldn’t really have an impact if it were to be used to process unstructured data (0.2) such as detecting major trauma events from wearables sensor data and signaling emergency services even though here is a high amount of data to work with (1.7).
(From theosz.medium.com )
Finished reading “10 Steps to Adopting Artificial Intelligence to Your Business” Article
Here’s a TLDR of what the steps in the article:
1) Get familiar with what modern AI can do. Take advantage of online resources to learn about AI and be familiar with many concepts.
2) Identify the problem that you want to solve with AI. There are many modern day issues that the world faces so look around at the news and other platforms for inspiration
3) Prioritize the financial value of your AI. Since you’re trying to start a business it’s important to focus on how the AI will bring in money for you.
4) Acknowledge your current capabilities of your business. Oftentimes, developing a new technology can take a long time so figure out what is currently possible and what is not.
5) Bring in Experts in AI and Set Up a Pilot Project (usually a first project). The scale of this project doesn’t need to be big. It should be finished within 2 to 3 months.
6) Form a group of people to clean your data. Although not mentioned in the article, I believe that this can be solved through open sourcing.
7) Make sure to start small! Big businesses have all started small so it makes no sense for you to start big.
8) Take into account the amount of computing that your AI will require. Make sure to have enough storage for it.
9) Incorporate AI to your daily routine.
10) Make sure to take in account many factors as you develop your AI. Things such as power failures, budget, and storage should be accounted for so make sure you have a plan to deal with these.
(From pcmag.com)
Finished reading “4 Business Applications for Natural Language Processing” article
The 4 ideas for NLP usage that the article mentions are as follows:
* Neural Machine Translation
* Chatbots
* Hiring Tools
* Conversational Search
(From cio.com)
Finished reading “6 Ways to Boost your Marketing for Natural Language Processing” article
The 6 ways to boost marketing for NLP are as follows:
* Give it more personality and more customer service capabilities
* Include voice search capabilities
* Include sentiment analysis for understanding customers. This way you can better serve their needs.
* Include automated summarization. This technology has grown throughout the past couple of years and has become more abstract with summarizes. These can be used for a variety of things and is a great factor for marketing
* Use AI to create a better slogan for your business
* Use AI Writer for better content generation. The newly generated content can be attractive aspect for new customers.
(From forbes.com)
For this week I wanted to try a few new things. Here’s a TLDR of my thought process:
* I’ve noticed that my progress has been slowing down for most of week 2 because I was spending so much time on the machine learning tutorial. It seems that trying to get everything to work is harder than I originally anticipated and it’s becoming a burden. I know my time here at Techie Youth is limited so I decided that I want to try out another course of my interest sometime this week. Not exactly sure when but should be sometime this week. I want to move on to other topics in the AI course as well and try to bounce back between the AI course and the other course of my choosing.
* I’ve also realized how much time writing blogs can take due to the fact that trying to look back and forth at my notes as I write what I learned that day is time consuming as well as distracting. There’s also times where writing a blog can be a bit difficult especially during days where I’m working short hours and there’s not really much to talk about but the fact that I was continuing with a tutorial. So I’ve decided to shift my blog publishing to the last day I work for the week. This way I can primarily focus on the material I’m trying to learn and hopefully provide more insightful blog posts.
Hope these things work out and you’ll hear from me soon :3
For my last day in my second week at Techie Youth I decided to do some reviewing today for all the material I covered in the past 8 days. I started at the notes I took from the Python Tutorials and all the way to the very beginning covering all the terms I’ve learned so far in the course. To end off my day I decided to move on to the Neural Networks section and read a little bit about it. Here’s what I learned:
Neural Networks
* These are computer systems created to do machine learning. They are modeled loosely after human brains and can consist of thousands to millions of interconnected nodes.
Today’s Neural Networks are organized into layers of nodes and data moves only in one direction throughout these nodes.
* To each incoming connection a node will assign a number known as a “weight”. So when the network is active, a node receives a different item/ number over each of its connections and multiplies it by the associated weight. After adding the resulting products together and getting a single number, the network will now check if the number is below or above a certain number. If the number is below, nothing is sent to the next layer of nodes. If the number is higher, then the data is sent to the next layer. This is repeated until it reaches the outer layer. During training, these weights will continue to be adjusted and corrected until data passed in is labeled more accurately.
* The concept of Neural Networks being structured like way is from modern Computer Scientists and the advent of GPUs (Graphics Processing Units) have propelled the continued use of such structure. Neural Networks only continue to grow in complexity as modern day hardware can support these systems.
(From news.mit.edu article)
Finally, today is the day where I’m going to build a real decently sized project that I can maybe put on my resume! I was excited to get everything setup so I could begin development as soon as possible but life sure has its fair share of slaps in the face! Upon trying to download the libraries I needed for this project I realized that my python file for my project was not able to access the downloaded libraries, specifically SciPy, which is probably one of the most important files I need. I did some digging around the internet and tinkered with my terminal to re-install the library but for some reason nothing was working. I then decided to try to install the other libraries such as sklearn which was also not installed and that was trouble too. After hours of tinkering and looking at the web I finally found out that I could just install Anaconda which is an open-source Python distribution specifically FOR data science and machine learning! WOW if only I had found this sooner! It has all the libraries I needed that come with the installation. So after a frustrating 4 ½ hours I downloaded Anaconda, searched up how to use it in my code editor and BOOM, my world is now brighter than before. I hope I don’t encounter more issues down the line but I’m completely prepared and I’ll do my damn hardest to not lose my mind ;3
Today I finally finished the Python Advanced Tutorial Section and I think that the time I spent working in these sections actually paid off. The knowledge from this last section helped me with the small project that I was following along with today. It's a very small scale project but by following the tutorial step by step, I generally understood how a future project could look and it seems quite daunting. I encountered a handful of issues, including one of the files I needed not importing correctly and I got “ValueErrors” from two things in my code not matching. I had to reread the steps many times and I tried my best to understand everything the tutorial was trying to show me. This was definitely a tough day! Here’s my notes:
Finished Python Advanced Tutorial :
Code Introspection
The ability to examine classes, functions, and keywords for what they are, what they know, and what they know/ what information is inside them
The “dir()” function seems to be a function that I would use often
There’s more functions so make sure to check the documentation
Closures
A function object that can remember values in enclosing scopes even if it is not present in memory.
A Nested function is a function inside another function and these functions can access values of the enclosing scope (basically the function that this function is inside of).
Using the “nonlocal” keyword we can use variables we defined inside the nested function outside in the outer function
Using closures means that we don’t have to keep using global variables and we can kind of hide data
Decorators
Allows for simple modification to callable objects like functions, methods, or classes.
There’s plenty of Decorators in Python documentation so be sure to check it out
Map, Filter, Reduce
These three things allow a programmer to write shorter code in terms of iterating through a list of elements
Only two are built in - Map and Filter - meanwhile Reduce needs to be imported
The “map()” function returns a map in the most recent version of Python and is a generator object. This can be converted into a list using the “list()” function.
Map has special parameters where it first takes a function that will be applied to each element and then as many other arguments as you want.
“zip()” function is used to create tuples using iterables
Unlike “map()”, “filter()” requires there to be one iterable. You will still have to enter a function much like “map()” but this function has to return a boolean value
“filter()” only returns elements that evaluate to be true by the function passed in
“reduce()” takes a function that has two arguments, an iterable, and an optional initial value. What this does is it first applies the function using the initial values and continues to do so until there is nothing else in the iterable.
Finished Reading Article on AI Algorithms
* Algorithms are essentially computer programs/ code that contains instructions for computers to accomplish a certain task
* Algorithms can help humans accomplish many things including recommendations for entertainment to help humans do repetitive tasks faster with higher accuracy.
(From thenextweb.com article)
Finished Reading Article on Starting A Machine Learning Project Tutorial
* Reminded of the steps to creating a basic
* Learned the meaning of Univariate linear regression (which is fancy term for a one variable problem)
* Followed steps to creating a basic project to predict the salary of a data scientist based on the years of experience
* Was given the code to write as well as a small dataset that could be used
* Spent time getting understanding the given code as well as experimenting and testing the code out myself. I made sure to leave comments after determining what each line of code did.
* Learned how to visualize given data set and apply various pandas functions to get information out of dataset
* Reviewed datasets: Training Dataset - used to train our model and Testing Dataset -used to test our model to see if it’s predictions are correct
* “ValueError” occurs when two pieces of data that are trying to be compared don't match one another. Make sure to check if two data types are the same before comparing.
(From sql.info article)
After trying to understand and practicing Regular Expressions for 2 hours, I gave up. It simply still has a hard time grasping the knowledge and usage around it. So today I decided to move on in the advanced Python tutorials and tried my best to practice and understand each and everyone of them. I also visited a blog today that talked about how to build an AI system. It gave steps and advice on how to do so. I made sure to take down all the important steps. So here’s my notes for the day:
Exception Handling
Exception is basically Python’s solution to errors. For example: an exception is thrown at the user when the user forgets to define a variable and tries to print it
There is a way to avoid exceptions by using a try/except block. This is useful when a programmer doesn’t want an error to stop a program.
The try will tell the program to execute code that comes after it and python throws an exception, the except block is there to catch the certain exception and tell python to execute some other code
Sets
Are lists with no duplicate elements inside. They’re extremely useful for calculating differences between two different containers.
Set as functions that come with it such as “.intersection()” - which determines which element(s) is present in both sets, “.difference()” - which determines which element(s) is present in a set but not both, “.symmetric_difference()” - which determines which element(s) appeared once between two sets, and “.union()” - which combines two sets to form one big set.
Serialization
Definition Here: refers to the process of converting a data object (e.g., Python objects, Tensorflow models) into a format that allows us to store or transmit the data and then recreate the object when needed using the reverse process of deserialization
Python comes built in with JSON libraries for encoding and decoding JSON. To use it, remember to import it.
There’s two different formats for JSON data, it’s either string or the object data structure. In Python, the object data structure consists of lists and dictionaries nested inside of each other. It allows for python methods to add, remove, search, and list elements from the data structure.
“loads” method is used to load JSON back to a data structure
“dumps” method is used to encode a data structure to JSON
“pickle” is a proprietary data serialization method supported by Python
Partial Functions
These can be created from Python’s functools library and are useful in specific situations
This tool allows us to call “partial()” insert our function of choice inside the parentheses then input default parameters. The last parameter that we need can be inputted elsewhere in our code. This makes testing a bit easier especially when we need to test for specific cases, we don’t need to keep typing the same inputs over and over again.
Simple Way to Build an AI System
A large data set is important for training AI
Building an AI has become more cheaper
Here are some steps for getting started:
* First:
Think about the issue
What is the desired outcome for solving said issue
AI is a tool so how can you use this tool to solve this issue?
* Second:
Now we need to prepare the data, which there are two types
Structured Data: data that has some consistency and pattern to it, such as a data set on customers that has the customer's name and phone number
Unstructured Data: data that has no consistency and has patterns that are harder to see. This can include chat messages.
After we get our data we need to clean it and sort them a certain way so that it is useable
* Third:
Now we can choose the algorithm
Recall back to the types of AI Machine learning Algorithms you have covered before.
Think about how this algorithm will perform for the problem at hand
* Fourth:
Train the Algorithm
Here we need feed our model our cleaned data
It is important here to establish model accuracy
* Fifth
Choose a language
Depending on the algorithm and the tools you need, different languages have their own advantages
Python is currently one of the most popular so look here before you look elsewhere.
* Sixth
Choose a Machine Learning Service Platform so that you don't have to worry about paying.
I’d say this is the most important part :p
(From : becominghuman.ai)
Quick comment:
Man I have to remember to upload these more often . . .
I keep on forgetting to upload these at the end of the day even though I finished writing them. I guess this will me something I need to work on throughout my time here at Techie Youth :3. Anyways here's what I did on Friday July 8th:
Today I decided to spend most of my work time reviewing all the material that I had gone through for my first week at Techie Youth. I went through a bunch of notes I took this week and specifically focused on Probabilistic Graphical Models since it was something I knew that I struggled to understand. After a bunch of googling I can safely say that I understand it a bit better and I might have to take statistics sometime because it’s quite interesting. I then spent a lot of time doing coding problems on w3resource today to sharpen up my Python skills a bit and installed Python on my brand new computer.
Here’s some things I picked up from this reading:
The scattering of online resources can make it hard for beginners to get started on a project so it's important to make sure that you can organize the resources you need and tackle something based on something you want to learn.
Online tutorials are great for understanding introductory parts of ML and DS(Data Science)
It's important to Grasp the Concepts of Machine Learning. To do so it's a good practice to look for tutorials that show what Algorithms do and buy searching how that certain algorithm works on google.
After basic knowledge, learning various other algorithms can be a next step. Learn from authors and other online sources and try to implement the algorithms on your own.
(From towardsdatascience.com blog)
Today I was finally able to move on to the next section, AI Technical Skills, which basically taught Python. As someone that has used Python before I found the first section on the website, “Learn The Basics”, to be a great refresher. I spent my time mainly on this section and then moved on to the second section: “Data Science Tutorials”. I found the exercises and information within these sections to be engaging and helped me learn new things that I didn’t know before. Without further ado this is what I learned in each section:
(From learnpython.org)
Learn The Basics:
* Hello World - Reviewed how to print information in python
* Variables and Types - Reviewed different data types in python which include strings, ints, and floats. Reviewed concatenation in Python and how you could only use the ‘+’ operator in print when printing the same data type.
* Lists - Reviewed how to use Python lists and some functions you can use with it such as appending/ adding values to a list
* Basic Operators - Reviewed how to do basic mathematical operations in Python such as subtracting, adding, dividing, and multiplying. I decided to challenge myself here and write different mathematical formulas such as the quadratic formula.
* String Formatting - Relearned how to do string formatting in Python and using tuples for string formatting.
* Basic String Operations - Reviewed different operations that could be used on strings such as len, count, index, upper, lower, and split. I learned how to use bracket operators to manipulate string outputs as well as using boolean operators such as startswith and endswith to check if a string starts or ends with certain characters.
* Conditions - Reviewed boolean operators such as “and”, “or”, “not”, “in”, and “is”
* Loops - Practiced and reviewed how to perform for loops and while loops in python. I decided to practice on my own and create a simple program to loop through a list and check for odd numbers and if a certain letter was inside each string in a list. I also made sure to practice using “break” and “continue”.
* Functions - Reviewed how to write functions in python. Practiced using loops in functions and using boolean operators.
* Classes and Objects - Reviewed how to create Class objects in python. Practiced by creating functions in a class and used __init__ for initializing a class.
* Dictionaries - Reviewed and practiced how to create and use Dictionaries. Learned how to delete an element in a dictionary using “del “ and “pop”.
* Modules and Packages - Reviewed modules and relearned how to access information within modules and packages (you do this by using import). Practiced using objects from modules and calling specific functions from modules
Data Science Tutorials:
* Numpy Arrays - Learned that Numpy Arrays are alternatives to lists in python and there are many operations that can be done to these arrays that make life easier. Ex: If you want to multiply all numbers in a list by 2 you can just multiply the Numpy Array by 2. Learned to use the “type” function to print out a class type of an object.
* Pandas Basics - Learned about how Pandas is a high level data manipulation tool and its key data structure is “DataFrame”. We can create a “DataFrame” by passing in a dictionary(use pandas.DataFrame(dict)) object or a csv file (use pandas.read_csv). When we try to print this, we get our information in a grid-like format. I learned that I can use the “brics.index()” to edit index values. You can also index a “DataFrame” by using brackets.
To Index through columns write name of column: file[[ “name”, “number”]]
To Index through rows write, input the rows you would like to see: file[0:4]
I also learned how to use “iloc” and “loc” to perform data selection operations.
(Made sure to practice thoroughly)
Today I decided to grasp as much beginner knowledge as possible about Artificial Intelligence Algorithms. It was definitely more challenging to understand than beginner material from my Day 1 but I continued to find the material quite interesting. This day consists of both the reviewing of material as well as the learning of new material to finish the Algorithms section of the course. Here’s a list of the things I was able to accomplish today:
(From learntocodewith.me blog)
(From SAS)
(From SAS)
In general the data that is accumulated includes all decisions made by the program to solve a problem. If the program chooses an action to solve a problem and is successful, the data from this action would be saved. When the program encounters a similar problem in the near future it can use the previously stored data from the previous problem to see if this can best solve its current problem.
CBR has what is known as the 4R methods which it uses to learn how to solve problems:
Retrieval:
Process of finding a case that is similar to the current situation or state. This relies on an ability to assess any two states based on their applicability.
Re-use:
Retrieve a case and propose it as a valid action to apply to the current situation.
Revision
A proposed action is evaluated through a series of metrics or simulations. This will determine whether it is practical to apply it to the case we are in.
Retention:
Applied after a successful execution by storing the result of this experience in memory. This is useful when a proposed solution is revised and now given its individual case.
Using this process CBR can learn and be used to predict outcomes such as predicting the outcome of a court case etc.
(From AI & Games, AI 101)
Direct Graphical Models (DGMS)
* DGMs are models that depict dependent relationships between variables]
* These models have a series of nodes with arrows from certain nodes pointing to other nodes depicting the relationship they can have with one another. Nodes that have arrows pointing to one another are called cycles (imagine 3 circles in a triangle formation labeled a, b, and c where a has an arrow that points to b, b has an arrow pointing to c, and c has an arrow pointing to a). Not all DGMs have cycles, these are called Directed Acyclic Graphs (DAGs).
* An example of DGMs are Bayesian Networks (BNs) which are DAGS with random variables representing observable or latent variables of the model
(I will be calling them MRFs like the blog writer does)
* Much like BNs, MRFs are used to describe dependencies between random variables using a graph.
* These models look similar to DGMs but don’t have any arrows. This DOES NOT mean they do not contain cycles. There are 3 properties I learned about these models: Pairwise Markov Property, Local Markov Property, and Global Markov Property each describing possible relationships variables in a model can have with one another.
* Triangulation is the procedure of turning MRFs to BNs
* Moralization is used when converting a BN to a MRF
(From towardsdatascience.com blog)