<?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 S P on Medium]]></title>
        <description><![CDATA[Stories by S P on Medium]]></description>
        <link>https://medium.com/@sp7242?source=rss-8bdb466c205d------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/0*5aA73u-HEX7qyfcN.jpg</url>
            <title>Stories by S P on Medium</title>
            <link>https://medium.com/@sp7242?source=rss-8bdb466c205d------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Fri, 20 Nov 2026 07:55:08 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@sp7242/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[Data Science Demystified-From Being School Boy to Data Scientist]]></title>
            <link>https://medium.com/@sp7242/data-science-demystified-from-being-school-boy-to-data-scientist-ee967e2a5703?source=rss-8bdb466c205d------2</link>
            <guid isPermaLink="false">https://medium.com/p/ee967e2a5703</guid>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[data-science]]></category>
            <category><![CDATA[deep-learning]]></category>
            <category><![CDATA[machine-learning]]></category>
            <dc:creator><![CDATA[S P]]></dc:creator>
            <pubDate>Sat, 29 Feb 2020 01:14:56 GMT</pubDate>
            <atom:updated>2020-03-02T02:33:47.206Z</atom:updated>
            <content:encoded><![CDATA[<h3>Data Science Demystified — From Being School Boy to Data Scientist</h3><h3>Acknowledgements</h3><blockquote>For the later part of this blog post (Data Science — The Formal Way), I have borrowed ideas from Foundations of Data Science course by <a href="https://padhai.onefourthlabs.in/">One Fourth Labs</a>.</blockquote><p>Data Science consists of two words and we’ll look into each of them one by one. We will start with science.</p><h3>What is science ?</h3><p>Move back to your school days. Remember you studied science. It consisted of three core subjects — physics, chemistry and biology. What is common among these three which makes them categorized into science. We will look into each of them</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*g0JVdP1uQ0oVYF4hjUZvBQ.png" /><figcaption>Science (Courtesy : <a href="https://www.alamy.com/blackboard-school-science-chemical-teacher-board-drawing-vector-chemistry-classroom-sketch-students-laboratory-chalk-drawings-image211542701.html">Alamy</a>)</figcaption></figure><p>Lets start with <strong>physics</strong>. Say we are talking about gravitation law. Remember it was given by Sir Issac Newton. Yes, and he is in fact one of those two which is highly admired in physics (other being Albert Einstein). Now recollect that gravitational force is given by <strong><em>F = GMm/r²</em></strong> where G is universal gravitational constant, M and m are masses of two objects and r is the distance between their centers.</p><p>How did he found this formula. Obviously he did experimentation. He must have carried out many experiments — taking different masses of different radii and so on and thus found out this expression and also the universal gravitational constant value too. This was one of the greatest discovery. And that’s what science is. You come with different facts by doing experiments on your subject of interest. These discoveries in turn ignite others.</p><p>Lets also talk about <strong>chemistry</strong>. Remember molecular structures, orbital theory (MOT), structure of atom etc. How did scientists came with these. Again, they did many experiments and proposed their own theory to explain those phenomena. These theories hold good for most or all of the cases and hence got widely accepted. Again it is same experimentation , i.e. trying out new and new things to know more about the interest, that triggered these theories. And of course this is science.</p><p>At last also go with <strong>biology</strong>. What about cell structure. Lets be specific and talk about neural cell or neuron. This has to do with deep learning too, of course there we use artificial neuron and here biological or the natural one — but the concepts remains same after all. Talking about cell (neuron) structure, scientists must again have dealt with same experimentation and came up with their structure. Microscope or other high power instrument must have been used too. And one last time, this is what science is. You discover new things or facts by trying out new things.</p><p><strong>Now then what is data science..</strong></p><p>Obviously doing science on data. That is deriving new insights from data by doing many different experiments. So you are given data and you do many experiments to find many new interesting facts or trends on it.</p><p><strong>What is data?</strong></p><p>This is one of most important part of the subject data science. Without data you cannot do anything in this domain. If there is no data you cannot do science on it. So there has to be data. Now what is this data. It is in fact mostly obtained from real world.</p><p>Data is any information on which you want to apply scientific, mathematical and/or algorithmic techniques to derive insights or facts. Data can be structured or unstructured or semi-structured too. In school, when you sat on bench, you must have paid attention to what teacher said. That is you gather audio data in your memory (mind). Similarly for text data from textbooks and so on.</p><h3>Data Science - The formal way</h3><p>Now, doing data science is a systematic process. A very well organised, precise and pure definition for data science is given by <a href="https://padhai.onefourthlabs.in/">One Fourth Labs</a>. According to them, Data Science consists of five tasks to be performed on data namely collect, store, process, describe and model. Each of these tasks have be performed in systematic and elaborative manner. Let us look into each of these in detail.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/865/1*GWKtnKN8CfQcroQlG_zGRw.png" /><figcaption>Data Science Tasks (Courtesy : <a href="https://padhai.onefourthlabs.in/">One Fourth Labs</a>)</figcaption></figure><ol><li><strong>Collecting Data</strong><br>Now to do data science, at very first we need data itself on which we can apply scientific techniques. This is precisely done in this step of collecting data. There arise three kinds of situations while collecting data — <br><strong>a)</strong> <strong><em>Data already exists in database</em></strong> — Here only data needs to be accessed by database query languages like SQL.<br><strong><em>b)</em></strong> <strong><em>Data exists but at multiple sources </em></strong>— In such cases data needs to be crawled and scraped. Python, Java programming skills can be used.<br><strong><em>c) Data is not available </em></strong>— In such situation, various experiments are carried out to collect data.</li><li><strong>Storing Data<br></strong>Now we have data but it needs to be stored. Three techniques are used for this purpose -<br><strong><em>a) Storing data in single database (Relational Database) </em></strong>— In this technique structured data is stored in relational database.<br><strong><em>b) Storing data at multiple databases (Data Warehouse) </em></strong>— Here structured data is stored in multiple databases within the same organisation. These databases are collectively called as data warehouse. <br><strong><em>c) Storing all forms of data in Data Lakes</em></strong> — Here all forms of data are stored — structured, unstructured, semi-structured are stored.</li><li><strong>Processing Data<br></strong>After collecting data, next step is to process the data. Stored data cannot be directly used for modelling as there might be issue with its format or values. Following methods comprise of data processing step :<br><strong><em>a) Data Munging or Wrangling </em></strong>— Here useful data is extracted, transformed to appropriate format and loaded into our own database. This is also known as Extract, Load and Transform (ETL) process.<br><strong><em>b) Data cleaning —</em></strong> Data may contain missing values, outliers (those data points which doesn’t correspond to our data) etc. All these are handled in this method.<br><strong><em>c) Data Scaling, Normalising and Standardising</em></strong> — <br>Data may need to be scaled to appropriate unit for our own database. Eg. Data may contain attribute in USD unit but You want to scale it to INR. In data normalisation, data is clamped between 0 and 1 (both inclusive). This is done in order to avoid variation in range of data values.<br>In data standardisation, data is transformed to zero mean and unit variance. This is done to get good sense of how various data are distributed and hence compare them.</li><li><strong>Describing Data<br></strong>We need to know about our data before modelling it. This is precisely done in this step. Data can be described in two ways :<br><strong><em>a) Data Visualisation </em></strong>— Here, graphical tools are used to plot data like bar graph, histogram, scatter plot etc.<br><strong><em>b) Data Summarisation</em></strong> — Here data is described in terms of mean, median, mode, variance, standard deviation, correlation etc of various attributes of data.<br>Describing data deals with descriptive statistics.</li><li><strong>Modelling Data<br></strong>This is the final step of data science process. Model can be thought of as function and is in fact that. More formally model is our approximation of data distribution. Here we try to find function which maps input(s) to output(s). Once we get that function, it can be used to find responses for other unknown inputs (that is whose data is not available).<br>Two approaches are used for this — <br><strong><em>a) Statistical modelling</em></strong> — Here we assume that our data came from some simple model like linear model. More specifically output variable(s) is some simple function over input(s). We then formulate our hypothesis and test it by giving some statistical guarantees like p-values, goodness of fit etc. Statistical Modelling deals with inferential statistics.<br><strong><em>b) Algorithmic modelling</em></strong> — This is also known as machine learning. Here we assume that our data distribution is modelled on some complex function. Of course data can also come from some simple distribution and in such case model may or may not adjust to this simple one. We just feed the data to our model and then estimate function using algorithms and optimization techniques. This function is generally complex, but as stated above it can be simple too.</li></ol><h3>Conclusion -</h3><p>In this post, I have explained the core concepts behind data science and related it to school science. We have looked at various tasks performed in this domain.</p><h3>References -</h3><p><a href="https://padhai.onefourthlabs.in/">https://padhai.onefourthlabs.in/</a><br><a href="https://www.statisticshowto.datasciencecentral.com/normalized/">https://www.statisticshowto.datasciencecentral.com/normalized/</a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=ee967e2a5703" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>