<?xml version="1.0" encoding="UTF-8"?><rss xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom" version="2.0" xmlns:cc="http://cyber.law.harvard.edu/rss/creativeCommonsRssModule.html">
    <channel>
        <title><![CDATA[Stories by Olaniyi Yusuf on Medium]]></title>
        <description><![CDATA[Stories by Olaniyi Yusuf on Medium]]></description>
        <link>https://medium.com/@olaniyiyusuf2000?source=rss-47f4e90fd4bf------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/0*E1xqp1tFoJXxHnuS.jpg</url>
            <title>Stories by Olaniyi Yusuf on Medium</title>
            <link>https://medium.com/@olaniyiyusuf2000?source=rss-47f4e90fd4bf------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Fri, 20 Nov 2026 16:14:49 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@olaniyiyusuf2000/feed" rel="self" type="application/rss+xml"/>
        <webMaster><![CDATA[yourfriends@medium.com]]></webMaster>
        <atom:link href="http://medium.superfeedr.com/" rel="hub"/>
        <item>
            <title><![CDATA[From Y=MX+C to Deep Neural Network — Part 2]]></title>
            <link>https://medium.com/@olaniyiyusuf2000/from-y-mx-c-to-deep-neural-network-part-2-429fa0cbbf97?source=rss-47f4e90fd4bf------2</link>
            <guid isPermaLink="false">https://medium.com/p/429fa0cbbf97</guid>
            <category><![CDATA[machine-learning]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[neural-networks]]></category>
            <category><![CDATA[gradient-descent]]></category>
            <category><![CDATA[deep-learning]]></category>
            <dc:creator><![CDATA[Olaniyi Yusuf]]></dc:creator>
            <pubDate>Mon, 15 May 2023 08:21:52 GMT</pubDate>
            <atom:updated>2023-05-15T08:26:27.770Z</atom:updated>
            <content:encoded><![CDATA[<h3>From Y=MX+C to Deep Neural Network — Part 2</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*NDqI_qlrL8UUTJOwE6dkuw.jpeg" /><figcaption>image from istockphoto.com</figcaption></figure><h3><strong>Table of Content</strong></h3><ol><li>Introduction.</li><li>Implementing Forward Propagation.</li><li>Backward Propagation and Gradient Descent.</li><li>Conclusion.</li></ol><h3><strong>Introduction</strong></h3><p>Welcome to the second part of this series on building a neural network from scratch using the Python NumPy library! If you haven’t had a chance to review the previous post, I encourage you to do so <a href="https://medium.com/@olaniyiyusuf2000/from-y-mx-c-to-deep-neural-network-part-1-562d87fb2b23">here</a>, as it covers the introductory concepts that form the basis for this implementation. Understanding those concepts will give you the necessary background knowledge to follow along and make the most of this tutorial. So, if you’re ready, let’s dive into the implementation process and continue our journey into building neural networks!</p><h3>Implementing Forward Propagation</h3><p>In forward propagation for a simple neural network with a single hidden unit, the following stages are involved:</p><ul><li>The weights and biases are randomly initialized at the input layer and passed to the hidden layer.</li><li>The value for y as wx+b is computed in the hidden layer and passed to the activation function to produce a non-linear output.</li><li>The output from the activation function is passed to the output layer, which produces the probability of the output as the prediction.</li></ul><p>It’s important to note that the activation function helps the neural network to identify non-linear relationships in the input data, allowing it to make more accurate predictions. Additionally, the weights and biases are adjusted during training using the backward propagation stage, which helps to optimize the network’s performance.</p><p><strong>Let’s implement forward propagation.</strong></p><p>For the purpose of simplicity, we will start with a single neuron</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/628/1*crjuKTtFQ7xrx9GtUOY9CQ.png" /><figcaption>image created with paint3D</figcaption></figure><p><strong>Notations</strong></p><ol><li>a[l] means a for the l-th layer.</li><li>a(n) means the n-th example in a and</li><li>a[l](n) is the n-th example in the l-th layer.</li></ol><p>Moving forward,</p><p>1. Get our weighted sum i.e z = wx +b</p><p>2. Apply the activation function to make it non-linear, we will be using a sigmoid function.</p><p>3. Use the output in 2 as an input in the next layer.</p><p><strong>Between layers 0 and 1:</strong></p><p>z[1] (1) = w[1] (1) * A[0] (1) + b[1] (A[0] (1) is X1)</p><p>A[1](1) = sigmoid(z[1](1))</p><p><strong>Between layers 1 and 2:</strong></p><p>z[2] (1) = w[2] (1) * A[1] (1) + b[2]</p><p>A[2](1) = sigmoid(z[2](1))</p><p>Building up on this, let’s consider a neural network shown below.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/553/1*_7XSrc19L11kJyxcPL2oMQ.png" /><figcaption>image created with paint3D</figcaption></figure><p><strong>Between layers 0 and 1:</strong></p><p>Weighted sums:</p><p>z[1] (1) = w[1] (1,1) * A[0] (1) + w[1](2,1) * A[0] (2) + b[1](1)</p><p>z[1] (2) = w[1](1,2) * A[0] (1) + w[1](2,2) * A[0] (2) + b[1](2)</p><p>z[1] (3) = w[1](1,3) * A[0] (1) + w[1](2,3) * A[0] (2) + b[1](3)</p><p>Activated Sums:</p><p>A[1](1) = sigmoid(z[1](1))</p><p>A[1](2) = sigmoid(z[1](2))</p><p>A[1](3) = sigmoid(z[1](3))</p><p>To implement this in Python, we have to use a for loop, however, with our knowledge of matrices, we can simplify this process using numpy arrays which require lesser complexity both in terms of time and space. You can read more about time and space complexity to understand why array is preffered.</p><p>So we will have:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/628/1*1cQfVZUQ_W7pQMHBHGNyug.png" /></figure><p>Therefore,</p><p>Z[1] = W[1] * A[0] + b[1] which will give us a matrix of shape 3,1</p><p>A[1] = sigmoid (Z[1]) — input for next layer</p><p>You can try the matrices multiplication. Check this <a href="https://www.youtube.com/watch?v=2spTnAiQg4M">video</a> on how</p><p>The general equation for weighted sum can be written as</p><p>Z[L] =W[L] * A[L-1] + b[L]</p><p>A[L] = sigmoid(Z[L])</p><p>And the shapes are:</p><p>W[L] = n,n-1</p><p>A[L-1] = n-1,1</p><p>b[L] = n,1</p><p>Z[L] = n,1</p><p>A[L] = n,1</p><p>Where n is the number of examples in laye rL and n-1 is the number of examples in layer L-1.</p><p>This process will then be repeated for all layers as shown in the code below using a for loop untill you get your y_pred.</p><h3>Backward Propagation and Gradient Descent.</h3><p>Remember that one of the major development data science has brought to various industries it’s applied is the ability to make data-driven decisions of which neural network is not an exemption.</p><p>In the case of a neural network, it will learn from its past mistakes to optimize the algorithm. This mistake is usually gotten from the cost function (How costly is this error? 😄)</p><p>This learning process is called backward propagation and the methodology used is called gradient descent.</p><p>Gradient descent is a powerful optimization algorithm that is widely used in machine learning. It works by finding the turning point (minimum point in this case) of a function. Here are the steps involved in applying gradient descent:</p><ul><li>First, we get the predicted value for a given input.</li><li>We then calculate the loss between the predicted value and the actual value using a loss function. In binary classification, a commonly used loss function is binary cross-entropy, which is given by L(a,y) = -ylog(a) -(1-y)log(1-a), where a is the predicted value and y is the actual value.</li><li>We differentiate the loss function with respect to the weight and bias, i.e. dL/dw and dL/db, respectively.</li><li>Finally, we update the weight and bias by subtracting the product of the learning rate and the differentials dL/dw and dL/db, respectively. This helps to optimize the network’s performance.</li></ul><p>So, in summary, gradient descent is an iterative process that helps to optimize the neural network’s performance by minimizing the loss function. By updating the weights and biases based on the gradients, we can iteratively improve the model’s accuracy.</p><p>Therefore, we will have:</p><p>W = W — learn_rate * dL/dw</p><p>b = b — learn_rate * dL/db</p><p>Now let’s differentiate to get dL/dw</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*uy6pNHILW8VvutSCvX7cWA.jpeg" /><figcaption>image from istockphoto.com</figcaption></figure><p>Recall from the forward propagation step,</p><p>Z =f(W,b)</p><p>A =g(Z) and</p><p>L = L(A,y)</p><p>To get dL/dw, we will be using chain rule which state that the derivative of a composite function is the product of the derivative of the outer function and the derivative of the inner function.</p><p>More formally, let f(x) and g(x) be two functions, where g(x) is the inner function and f(g(x)) is the outer function. Then, the chain rule states that the derivative of the composite function f(g(x)) is given by:</p><p>(d/dx) [f(g(x))] = (df/dg) * (dg/dx)</p><p>So in this scenario, assuming b is 0, we can say</p><p>Z= f(w)</p><p>a = g(Z) = g(f(w))</p><p>L = L(a) = L(g(f(w))). Therefore,</p><p>dL/dw = dL/da * da/dZ * dZ/dw</p><p>Recall, the differential of log(x) is 1/x</p><p>So, for L = -yloga — (1-y)log(a)</p><p>We would have dL/da = -y/a + (1-y)/(1-a) ( y is a constant like 2 in this case)</p><p>Just to understand how the differential of (1-y)log(1-a),</p><p>Say let 1-a be c and b =(1-y)log(c)</p><p>using the chain rule,</p><p>db/da = db/dc * dc/da.</p><p>db/dc =(1-y)/c and dc/da = -1</p><p>Therefore db/da = (1-y)/c * -1</p><p>db/da = -(1-y)/1-a</p><p>so when you multiply db/da above by the negative sign, we will have + (1-y)/(1-a)</p><p>To get da/dz:</p><p>Recall that a is gotten by applying activation function on z. Let’s use a sigmoid function, a(z) = 1/(1+e-z)</p><p>Let h = e-z</p><p>k = 1+h</p><p>and a = 1/k = k^ -1 = f(k)</p><p>Therefore, da/dz = da/dk * dk/dh * dh/dz</p><p>da/dk = -k^-2 = -1/k² and</p><p>dk/dh = 1</p><p>dh/dz = -e^-z</p><p>Therefore, da/dz = -1/k² * 1 * — e^-z</p><p>da/dz = e^-z/(1+h)²</p><p>da/dz = e^-z/(1+ e-z)² = 1/(1+e^-z) * e^-z/(1+e^-z) = a * e^-z/(1+e^-z)</p><p>But e^-z/(1+e^-z ) can be written as</p><p>(-1+1 + e^-z)/(1+e^-z) = -1/(1+e^-z) + (1+e-z)/(1+e^-z)</p><p>Recall, a = 1/(1+e), therefore, e-z/(1+e-z) = -a + 1 = 1-a</p><p>So, da/dz = a(1-a)</p><p>For simplicity, dL/dw will be denoted as dw and dl/dz = dl/da * da/dz will be denoted as dz.</p><p>Therefore, dz = dL/da * da/dz</p><p>dz = (-y/a + (1-y)/(1-a)) * a(1-a)</p><p>Upon expansion,</p><p>dz = a-y</p><p><strong>Note</strong>: The value for dz changes depending on the activation function used.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*mZJSIX09Q5ZWY3DJPkxGSQ.jpeg" /><figcaption>image from istockphoto.com</figcaption></figure><p>Now let’s implement,</p><p>Recall from forward propagation that Z[L] = w[L] * A[L-1] assuming b[L] equals 0.</p><p>So in backward propagation dw[L] = dz[L] * A[L-1].T where A.T means the transpose of matrix A.</p><p>You can check click <a href="https://github.com/Yusuf-Olaniyi/Let-s_Build_A_Deep_NN/tree/main">here </a>to see how I built a deep neural network from scratch.</p><h3>Conclusion</h3><p>I would like to emphasize that the knowledge and insights shared in this series are based on the valuable teachings of Andrew NG’s Deep Learning Specialization, the first course available on Coursera. If you’re interested in delving deeper into the subject, I highly recommend taking the <a href="https://www.coursera.org/specializations/deep-learning#courses">course</a> as it offers a comprehensive and insightful learning experience.</p><p>To further enhance your understanding, I have also provided a notebook where you can explore a practical implementation of the concepts discussed in this series. You can access the notebook <a href="https://github.com/Yusuf-Olaniyi/Let-s_Build_A_Deep_NN/tree/main">here</a>. You can also access the previous post here.</p><p>Thank you for taking the time to read through this series. I appreciate your support, and I would love to hear your feedback. Please feel free to like, comment, share, and follow me for more informative content. You can also connect with me via <a href="https://www.linkedin.com/in/yusuf-olaniyi/">LinkedIn</a>, <a href="https://twitter.com/Yusuf_Olaniyi1">Twitter</a>, or Email me at olaniyiyusuf2000@gmail.com for job recommendations, projects, and potential collaborations.</p><p>Remember, the world of neural networks is vast and ever-evolving, and I look forward to witnessing your contributions in this exciting field. Happy exploring!</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=429fa0cbbf97" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[From Y=MX+C to Deep Neural Network — Part 1.]]></title>
            <link>https://medium.com/@olaniyiyusuf2000/from-y-mx-c-to-deep-neural-network-part-1-562d87fb2b23?source=rss-47f4e90fd4bf------2</link>
            <guid isPermaLink="false">https://medium.com/p/562d87fb2b23</guid>
            <category><![CDATA[data-science]]></category>
            <category><![CDATA[deep-learning]]></category>
            <category><![CDATA[neural-networks]]></category>
            <category><![CDATA[machine-learning]]></category>
            <category><![CDATA[technology]]></category>
            <dc:creator><![CDATA[Olaniyi Yusuf]]></dc:creator>
            <pubDate>Mon, 15 May 2023 08:18:11 GMT</pubDate>
            <atom:updated>2023-05-15T08:22:57.956Z</atom:updated>
            <content:encoded><![CDATA[<h3>From Y=MX+C to Deep Neural Network — Part 1.</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/640/1*G5t1HelozwfylJmWojZY_w.jpeg" /><figcaption>image from pixabay.com</figcaption></figure><h3><strong>Table of Content</strong></h3><ol><li>Introduction.</li><li>The Human Brain and Neural Network.</li><li>Steps in Building a Neural Network.</li><li>What is an Activation Function?</li><li>Conclusion.</li></ol><h3>Introduction</h3><p>Have you ever wondered how machines can learn to perform complex tasks, like recognizing faces or predicting stock prices? It all starts with a basic mathematical concept — the equation of a straight line. While this may seem daunting, don’t worry — even if you’re not a math whiz, you can still understand the basics of building a machine learning model. In this write-up series, we’ll break down the process in a way that’s easy to follow and show you how you can get started with deep neural networks. Whether you’re a complete beginner or have some experience with machine learning, this article will help you gain a better understanding of this exciting field.</p><h3><strong>The Human Brain and Neural Network</strong></h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/640/1*inOYV8RVBYhjtN26h3WFoQ.png" /><figcaption>A neuron in the human brain from pixabay</figcaption></figure><blockquote><em>“I’m not trying to make a model of how the brain works. I’m looking at the brain and saying, ‘This thing works, and if we want to make something else that works, we should sort of look to it for inspiration.’” — </em>GEOFFREY HINTON</blockquote><p>According to the pioneer of neural networks, Geoffrey Hinton, neural networks are inspired by the way the human brain is organized, they are not exact copies of it. While the human brain is much more complicated, neural networks are made to process information by connecting layers, much like how neurons do. It’s important to keep in mind that not all neural networks are built in the same way, and some, like Convolutional Neural Networks (CNNs), are created to resemble other biological structures, such as the eyes. A neural network’s layers include:</p><ul><li><strong>Input layer</strong>: A pass-through layer that accepts data for processing.</li><li><strong>Hidden layer(s)</strong>: Layers that analyze and identify patterns in the input data, much like how the human brain processes information. Each hidden layer can detect increasingly complex features of the input data.</li><li><strong>Output layer</strong>: The final layer of a neural network, which produces a prediction based on the input and the patterns identified by the hidden layers.</li></ul><p>This layered structure allows neural networks to perform complex tasks such as image recognition and language processing.</p><h3><strong>Steps in Building a Neural Network</strong></h3><p>When training a neural network, two main stages are involved: forward propagation and backward propagation.</p><p>Forward propagation involves processing input data through the network’s layers, including computing a weighted sum of the inputs and adding a bias term. This computation can be represented by the linear equation y = wx + b, where “w” represents the weights, “x” represents the input, and “b” represents the bias. As the input data moves through the layers, the network identifies patterns in the data and makes a prediction or classification based on those patterns.</p><p>Backward propagation, also known as backpropagation, involves adjusting the weights and biases based on the error between the predicted output and the actual output. This error is calculated using a loss function, which is a measure of how well the network is performing on a given task. By computing the gradients of the loss function concerning the weights and biases, the network can adjust these parameters to improve its accuracy over time.</p><p>These two stages are essential for training a neural network and getting it to perform well on specific tasks such as image recognition or language processing.</p><p><strong>Let’s deep dive into Forward Propagation:</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/280/1*9HOVUG3fbcibWTU9kM7_tg.jpeg" /></figure><p>Recall from the equation of the straight line (y=mx+c) which implies that y is dependent on x i.e. if x changes y will also change but the extent to which y will change for the same value of x is dependent on the gradient m. For example, consider the lines below</p><p>y = 4x+2 and y=2x+2</p><p>At x=5, the first line will return y as 22 while the second will return it as 12 which shows the impact of the gradient (i.e dy/dx in calculus and m in the equation of straight line)</p><p>In neural networks, our gradient(m) and y-intercept(c) are replaced with weight(w) and bias(b) respectively. To understand this better, biases are additional constants that must be added to the value of the output in the layer; these are pre-defined conditions in real life while weight is the rate at which the output value(y) changes concerning the input value(x). Therefore our first equation becomes y = wx+b.</p><h3><strong>What is an Activation Function?</strong></h3><p>In machine learning, problems are generally divided into two major categories: classification and regression. A classification problem requires the output to be true or false, while a regression problem requires the output to be a numerical value. However, the output of the formula y = wx + b is typically a numerical value, which is not ideal for a true or false prediction.</p><p>An activation function is a mathematical function used to convert a linear input to a non-linear output. Activation functions are used not only in converting linear outputs to non-linear but also in improving the performance of the model. Various types of activation functions are available, each with unique properties and advantages. Here are some of the most commonly used activation functions:</p><ul><li><strong>Sigmoid function</strong>: This function returns a value between 0 and 1, making it ideal for binary classification problems. The equation for the sigmoid function is: σ(x) = 1 / (1 + e^-x)</li><li><strong>ReLU (Rectified Linear Unit)</strong>: This function sets all negative values to zero and leaves all positive values unchanged. This is one of the most popular activation functions and is commonly used in deep learning. The equation for ReLU is: f(x) = max(0, x)</li><li><strong>Tanh (Hyperbolic Tangent)</strong>: This function returns a value between -1 and 1, making it ideal for classification problems that require more than two classes. The equation for Tanh is: tanh(x) = (e^x — e^-x) / (e^x + e^-x)</li><li><strong>Leaky ReLU</strong>: This function is similar to ReLU, but it allows a small, non-zero gradient for negative input values, preventing the “dying ReLU” problem. The equation for Leaky ReLU is: f(x) = max(αx, x), where α is a small constant (e.g., 0.01)</li></ul><h3>Conclusion</h3><p>In the concluding part of this write-up series, which you can find <a href="https://medium.com/@olaniyiyusuf2000/from-y-mx-c-to-deep-neural-network-part-2-429fa0cbbf97">here</a>, I explained the implementation process of building a neural network from scratch using the Python NumPy library. To further enhance your understanding, I have also provided a notebook where you can explore a practical implementation of the concepts discussed in this series. You can access the notebook <a href="https://github.com/Yusuf-Olaniyi/Let-s_Build_A_Deep_NN/tree/main">here</a>.</p><p>Thank you for taking the time to read through this series. I appreciate your support, and I would love to hear your feedback. If you found it helpful, please consider liking, commenting, sharing, and following me for more insightful content. Feel free to connect with me via <a href="https://www.linkedin.com/in/yusuf-olaniyi/">LinkedIn</a> , <a href="https://twitter.com/Yusuf_Olaniyi1">Twitter</a> or Email me at olaniyiyusuf2000@gmail.comfor job recommendations, projects, or potential collaborations. I’m excited to continue this learning journey with you!</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=562d87fb2b23" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[How to Launch your Tech Career as an A-student.]]></title>
            <link>https://medium.com/@olaniyiyusuf2000/how-to-launch-your-tech-career-as-an-a-student-4b66ed250947?source=rss-47f4e90fd4bf------2</link>
            <guid isPermaLink="false">https://medium.com/p/4b66ed250947</guid>
            <category><![CDATA[students]]></category>
            <category><![CDATA[personal-development]]></category>
            <category><![CDATA[technology]]></category>
            <category><![CDATA[learning-to-code]]></category>
            <category><![CDATA[tech-career]]></category>
            <dc:creator><![CDATA[Olaniyi Yusuf]]></dc:creator>
            <pubDate>Wed, 01 Mar 2023 15:06:31 GMT</pubDate>
            <atom:updated>2023-03-01T15:06:31.874Z</atom:updated>
            <content:encoded><![CDATA[<h3>Introduction</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*b8kRMPkxnQm0fR74QOFcZA.jpeg" /><figcaption>istockphoto.com</figcaption></figure><p>Every A-grade student, or any student who desires to be an A-grade student, may experience difficulty learning a skill and hands-on experience outside the typical school activities. Like any A-student, I struggled with this for the first three years of college. After some trials, though, I was able to get over the barrier, so I’ll be sharing some of the most important lessons I learned.</p><p>I will also provide my academic record to give you a sense of the type of student I am, how my everyday routines were before I started learning a skill, and how my daily activities are now that I started acquiring skills.</p><h3>My Academic Background and Turning Point.</h3><p>I am a fourth-year civil engineering student at the University of Lagos, ranking top 1% in out of 55000+ students in my school and second in my class. To better understand what this means, we had over 20 first classes after our first semester and I was not among the first 10 students. However, due to the demands of school and the desire to learn new skills, we all have to decide whether to prioritize school over learning a skill or the other way around.</p><p>Like every student, I also have to pick one, you can guess what I chose right? Okay, You guessed right, I chose the school. So for my first three sessions as a student, I prioritized getting an A in all my courses every semester, although I did not achieve this but I had an A in over 90% of the courses I have done so far.</p><p>I began to question why we have to choose between schoolwork or learning a new skill after committing a significant portion of my student life to it. Why can’t we go for both? And that was the defining moment for me. I’m assuming you are aware of the reasons why A-students who wish to engage in a field-related job can find it difficult to choose one. (What are your thoughts on this?)</p><h3>The Journey and Lesson</h3><p>At the beginning of 2022, I decided to learn data science and I focused on machine learning to be more specific. To be quite honest, there were many excellent and unpleasant memories along the way, but these are the things that really helped.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*46kX5tXSeATjcovWn9-Dyg.jpeg" /><figcaption>istockphoto.com</figcaption></figure><ol><li><strong>Start with Why</strong>: It’s crucial to have a valid motivation for learning the skill. I’m sorry to say it, but if making money is your primary motivation for learning a tech skill, you might be let down. Don’t get me wrong, money is important however, most tech bro and sis won’t tell you that they did a lot of free work or got under paid at the beginning of their career. Therefore, you need something to keep you motivated so that no matter what happens, you can still learn. You can read more about the significance of this point in Simon Sinek’s book — <a href="https://simonsinek.com/books/start-with-why/">Start with Wh</a>y</li></ol><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*WZ39ddAX9uVYJuznP6qTmQ.jpeg" /><figcaption>istockphoto.com</figcaption></figure><p><strong>2. Goal Setting</strong>: Setting SMART goals at regular intervals is crucial. Rather than setting an annual goal, I make quarterly goals which are then broken down into monthly, weekly, and daily goals. I then regularly assess these goals to determine how close I am to my overall goal. You can check <a href="https://medium.com/@ZaynabAwofeso/5-tips-that-helped-me-be-more-productive-in-the-first-quarter-of-2022-f7c53cec3a97">this</a> post on goal setting. Here are few examples of SMART goals:</p><ul><li>Study the first chapter of Data science workshop textbook by 12 noon.</li><li>Write about 5 data types and share on Analytics Vidhya by Friday.</li><li>Create my data science portfolio with an EDA project, Linear regression project, Case study project and Deep learning project before the end of April, 2023.</li></ul><p>3. <strong>Planning</strong>: Proper prior planning prevent poor performance, aside from setting goals, it’s important one form an habit of planning properly to help you manage your time and energy efficiently. This is one attribute that distinguish high archievers from their peers because they manage their time and energy. Like a friend of mine will say, “<em>Time and energy are finite resources</em>” .</p><p>In managing your time and energy, various people use different techniques, and I think having a personal timetable should work for most students because our school activities usually reoccur at regular interval and thier are usually a lot of distractions.</p><p>Here are some tips on forming an healthy and effective personal timetable that can help you manage your daily activities.</p><ul><li>Ensure you give yourself time for compulsory school activities.</li><li>Take breaks in between activities.</li><li>Take out time to do something you care about.</li><li>Leave out a day or two(maybe weekends) to carry out activities you missed during the week.</li><li>Make your time table flexible.</li><li>Add time to sleep and take some nap to regain energy.</li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*R3rC19cec_rezUgLehWnDg.jpeg" /><figcaption>istockphoto.com</figcaption></figure><p>4. <strong>Accountability Partner</strong>: Just setting a goal and failing to meet it is insufficient. In order to ensure that you are making progress, I therefore advise finding an accountability partner with a similar level of commitment.</p><p>I strongly advice that your accountability partner should be a friend, family member or an elderly person you respect and with a similar goal and try as much as possible to make the relationship fun and enjoyable so you won’t get tired of the person.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*lcttDP6myfh50H2fDgBRQA.jpeg" /><figcaption>istockphoto.com</figcaption></figure><p>5.<strong> Projects</strong>: As the saying goes, ‘you learn by doing’. Tech skills are not exempted in any manner, and working on projects is the greatest approach to develop your talents rather than enrolling in endless classes. Don’t get me wrong, it’s good to take courses, but be sure to work on a project relating to the course you completed after each one. Therefore, I recommend you check out world quant university project based data science course <a href="https://www.wqu.edu/">here</a> if you are interested in data science.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*zxJou5ue1-7rEk7STZrDPA.jpeg" /><figcaption>istockphoto.com</figcaption></figure><p>5. <strong>Ask the Right Questions</strong>: Yes, Google and other search engines are your friends, but it’s important to remember that there may be some unique ways to advance your career in your community or industry, so checking Google might not offer the best solution. Instead, ask people in your community who are already successful in the field you want to enter for advice that will help you.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*pe-2zhrFk5Dye-4Ur4Hz3Q.jpeg" /><figcaption>istockphoto.com</figcaption></figure><p>6. <strong>Community</strong>: It’s been said that two good heads are better than one. In order to learn about possibilities, receive guidance, collaborate, and give back, it’s critical to join a community of like-minded individuals in your school or neighborhood.</p><h3>Conclusion</h3><p>Discipline is one attitude that all A students share. And without discipline, all of this might not lead to good results, therefore exercise this with discipline.</p><p>Kindly share your opinions on the write up and the mindset that has guided you through your tech career to date.</p><p>Most importantly, I would appreciate if you can share this write up with your friends or colleagues that needs this.</p><p>Thank you for reading till the end, I am open to work on Data Analysis/Science and machine learning projects in the finance, health or education sector and similar opportunities too, you can reach out to me via <a href="https://www.linkedin.com/in/yusuf-olaniyi/">LinkedIn</a> or <a href="https://twitter.com/Yusuf_Olaniyi1">Twitter.</a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=4b66ed250947" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[How did the richest man in Babylon build his wealth?]]></title>
            <link>https://medium.com/@olaniyiyusuf2000/how-the-richest-man-in-babylon-build-his-wealth-a3c3305dd844?source=rss-47f4e90fd4bf------2</link>
            <guid isPermaLink="false">https://medium.com/p/a3c3305dd844</guid>
            <category><![CDATA[financial-planning]]></category>
            <category><![CDATA[personal-development]]></category>
            <category><![CDATA[wealth]]></category>
            <category><![CDATA[wealth-creation]]></category>
            <category><![CDATA[personal-finance]]></category>
            <dc:creator><![CDATA[Olaniyi Yusuf]]></dc:creator>
            <pubDate>Sat, 21 Jan 2023 20:38:18 GMT</pubDate>
            <atom:updated>2023-01-21T22:14:20.660Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/612/1*Atzt8d5tWe1mb269hxSRLA.jpeg" /><figcaption>istock.photo.com</figcaption></figure><h3><strong>Introduction</strong></h3><p>Every Gen Z that seeks advice about money, career, and life generally in a way should have heard the quote that ‘Everything is not about money’, however, our elders will agree with me that, without money, some basic things won’t be available. These two ideas seem contradictory but I think we can say, &quot;Even if everything is not about money, money is about everything.”(I would love your opinion on this in the comment section). So out of curiosity, I started reading books about money and some history of rich people in ancient days. So, I started this year with The Richest man in Babylon by George S. Clason. It’s important to note that, the book was first published in 1962, and it’s referred to as the success secrets of the ancient.<br>And I will be sharing the key things I learned from the book and my opinion about some of the concepts.</p><h3>Summary of the stories in the book.</h3><p>The Richest man in Babylon was a man called Arkad and Babylon was known by all and sundry as a city of the rich. So Arkad was the Elon Musk of his time. While going through all the stories shared in the book, I realized that there are three significant steps to accumulating wealth. <br>1. Saving (Wealth Accumulation)<br>2. Investing (Wealth multiplication) and<br>3. Taking the right advice (Wealth maintenance).</p><p><strong>Saving:</strong> One thing all the stories have in common is that every successful person in the past started by saving some percentage of their income before spending it on any other thing. The stories showed that 10% is a great way to start with. I particularly like this idea because human wants(desires) are unlimited and if a person cannot suppress his/her desire to suit what he/she has, such a person will end up in long-term debt. Just as the saying goes, cut your coat according to your cloth(not size).</p><p><strong>Investing: </strong>the same idea was shared in Rich dad poor dad by Robert Kyosaki, he termed this as making your money work for you. After saving 10% of your income constantly, the value of your savings needs to constantly increase in its ability or capacity and the best way to do this is by investing your savings wisely. The question now is, how does one identify a wise investment? Let’s go to the last point.</p><p><strong>Taking advice:</strong> If one invests his/her savings carelessly, there is a high chance that such savings will be lost within a short period. This is why it’s important to invest in that which one is aware of or seek advice from the right people. A very common example I have seen around me is those investing in cryptocurrency without any knowledge of how to increase their earnings.</p><h3>Conclusion</h3><p>After completing this book, I think that people mistake building wealth as building rich because people think of wealth in terms of money that they have, however, wealth is a measure of both money and your potential to make more money.</p><p>I would love your feedback about the book, your thought about wealth accumulation, and any other feedback/questions.</p><p>Thank you. You can reach out to me via <a href="https://www.linkedin.com/in/yusuf-olaniyi/">LinkedIn</a> or send me an email at olaniyiyusufofficial@gmail.com.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=a3c3305dd844" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[DATA WRANGLING BASICS EXPLAINED WITH EXAMPLES]]></title>
            <link>https://medium.com/@olaniyiyusuf2000/data-wrangling-basics-explained-with-examples-ebe25c1ff4b9?source=rss-47f4e90fd4bf------2</link>
            <guid isPermaLink="false">https://medium.com/p/ebe25c1ff4b9</guid>
            <category><![CDATA[data-analysis]]></category>
            <category><![CDATA[data-wrangling]]></category>
            <category><![CDATA[data-science]]></category>
            <category><![CDATA[data-cleaning]]></category>
            <category><![CDATA[machine-learning]]></category>
            <dc:creator><![CDATA[Olaniyi Yusuf]]></dc:creator>
            <pubDate>Wed, 27 Jul 2022 19:48:11 GMT</pubDate>
            <atom:updated>2023-05-10T06:52:11.636Z</atom:updated>
            <content:encoded><![CDATA[<h3>Data Wrangling Basics Explained With Examples</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/800/0*lWscjVvnsSTL-6Kq" /><figcaption>image from pinterest.com</figcaption></figure><p><strong>TABLE OF CONTENT</strong></p><p>1. Introduction</p><p>2. Data Gathering</p><p>3. Data Assessment</p><p>4. Data Cleaning</p><p><strong>INTRODUCTION</strong></p><p>Data is said to be the fuel of the future and like every other fuel, it has to pass through the refinery to make it useful. We will be learning how to refine data to make it useful. The process of refining is not data cleaning. Don’t be surprised, it’s called data wrangling. Although, the misconception is that data cleaning is a major step in data wrangling but without the first two processes we can effectively clean data.</p><p>To give you a smooth ride, I will be comparing the process to how Petrol is gotten from crude oil. As expected the first step will be to extract the crude oil and then identify the impurities before we finally separate our end product, petrol.</p><p>Similarly in data wrangling, the stages are</p><p>1.<strong><em> Data Gathering: </em></strong>where we get the data from the source.</p><p>2. <strong><em>Data Assessment: </em></strong>where we check issues in the data.</p><p>3. <strong><em>Data Cleaning:</em></strong> where we solve the issues identified while assessing the data.</p><p><strong>DATA GATHERING</strong></p><p>Data gathering is the process of collecting data from its source. Data can be classified as primary or secondary data based on its purpose of collection in comparison with what you are using it for. Primary data is collected and used for the same purpose as NetFlix using customer data to provide movies recommendation. In contrast, secondary data is collected for a different purpose for which you are using it like when a business is using data acquired by the Government or another entity to make business decisions. Primary data can be collected using various methods depending on the organization while secondary can be collected by any of the listed methods although not limited to them.</p><p><strong>1.<em> Online download:</em></strong> This involves downloading data directly from where it’s saved online like downloading a dataset from Kaggle or any open data source. It’s as simple as clicking on the download button.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/800/0*oYAeqRIMXqFSMtqw" /><figcaption>image from pinterest.com</figcaption></figure><p>2. <strong><em>Web scrapping:</em></strong> this involves getting data directly from a data using libraries like Beautiful Soup, Selenium, or Autoscraper.</p><p>Code example using requests and Beautifulsoup.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/800/0*L5KVuEjnT7U0yMXT" /></figure><p>The request like is used to seek permission to get the contents on a particular site. Also, functions, like find or find_all, are used to get certain content from the website like</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/800/0*fAh6flONaGBNbLTx" /></figure><p>3. <strong><em>Application Program Interface(API):</em></strong> This is similar to web scraping but involves getting permission directly from the platform owner. Also, if datasets are changed, it’s automatically updated while in the case of web scrapping, you have to go through the process again.</p><p><strong>DATA ASSESSMENT</strong></p><p>Data assessment is the phase where data personnel carries out the necessary test on the data to know the issues in the data just like when a doctor carries out a test on a patient. Unlike humans, data can only be affected by two broad issues/ailments, which are:</p><p><strong><em>1. Quality issue</em></strong></p><p><strong><em>2. Tidiness issue</em></strong></p><p>Before diving into how to handle the issues faced by our data let’s understand what the issues are all about. What do you think will be the difference between a quality and tidiness issue. take a guess and I will like to know what your thought is here in the comment section.</p><p>Here is an excellent way to decode it, data having a quality issue is said to be dirty while one having a tidiness issue is said to be messy. To have a better understanding, let’s consider things that make the room in the picture dirty and messy.</p><figure><img alt="a messy room" src="https://cdn-images-1.medium.com/max/563/0*KfH24IxlHJfTKICw.jpg" /><figcaption>image from pinterest.com</figcaption></figure><p>Things that make the room dirty are content that doesn’t belong to the bedroom like bottles and cups while things that make it messy are things that belong to the room but are in the wrong positions like clothes on the floor, books, and cream by the side of the bed. Let’s use this same idea in our data.</p><p><strong>Data Quality Issue:</strong></p><p>Quality issues deal with the content of the data. There are four major quality issues.</p><p>1. <strong><em>Completeness:</em></strong> Presence of missing value</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/696/0*naGHOMnuyiAQManb" /></figure><p>2. <strong><em>Validity:</em></strong> Availability of erroneous input data or datatype. For example, having postal code as int datatype instead of string or date as a string instead of a DateTime datatype.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/381/0*0dgbVcKfC1Cg87aN" /></figure><p>3. <strong><em>Accuracy:</em></strong> Inaccurate data from the data source. For example, if the age of a student was given as 1000, 1000 is a valid data type but is it possible we have a student whose age is 1000.</p><p>4. <strong><em>Consistency:</em></strong> Inconsistent data can be valid and accurate but has various forms to represent the same entry. For example, USA, US, United States, and United States of America represent the same country. This is very common in proper nouns from my observation. The state column in the table below has GA</p><p><strong>Data Tidiness Issue:</strong></p><p>These issues have to deal with the structure of the inputs. There are three essential requirements to make data tidy.</p><p>1. Each variable forms a column.</p><p>2. Each observation forms a row.</p><p>3. Each type of observational unit forms a table.</p><p>A common example is having multiple entries in a single column. Like having a phone number and email address in the contact column which should be in different columns as phone number and email address respectively as seen in the table above.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/412/0*N6VDYFsIi7Oq4O44" /></figure><p>After identifying all the issues with your data, it’s important to document both the quality and tidiness issues observed in the data as shown below.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/800/0*m2ryIrzJyDTNCo4d" /></figure><p><strong>DATA CLEANING</strong></p><p>This simply involves solving the issues identified while assessing the data. This can be done manually or programmatically. For a small dataset with a few issues we can solve the issues manually from the source while for a large dataset, one should use the programmatic approach (Python built-in functions like Pandas, Numpy, e.t.c will be very useful here).</p><p>Irrespective of the framework you decide to use, it’s important to create a copy of the document before cleaning with the copy function(df_copy = df.copy()). Although it is not compulsory you use this but the DCT framework makes it easy for you and your fellow programmer to understand the why and how of each cleaning step.</p><p><strong><em>DCT Framework</em></strong></p><p><strong><em>D — Define the issue</em></strong></p><p><strong><em>C — Code</em></strong></p><p><strong><em>T — Test to confirm if the challenge has been solved.</em></strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/800/0*7dZiUxfCCluklPXB" /></figure><p>This was part of the few things I learned from taking <a href="https://www.udacity.com/course/data-analyst-nanodegree--nd002">Udacity Data Analyst Nanodegree</a> and you can click <a href="https://github.com/Yusuf-Olaniyi/weratedog_Data_Wrangling">here</a> to have access to one of my projects where I wrangled data using tweepy and Twitter API.</p><p>Kindly like, comment and share. You can reach out to me via <a href="https://www.linkedin.com/in/yusuf-olaniyi">LinkedIn </a>or <a href="/service/mailto:olaniyiyusuf2000@gmail.com">Email</a>.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=ebe25c1ff4b9" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>