<?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 Gunjan Grunge on Medium]]></title>
        <description><![CDATA[Stories by Gunjan Grunge on Medium]]></description>
        <link>https://medium.com/@stdevilgunjan?source=rss-cca56ab38963------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/0*acKNtGUAAdlCg6xL</url>
            <title>Stories by Gunjan Grunge on Medium</title>
            <link>https://medium.com/@stdevilgunjan?source=rss-cca56ab38963------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Thu, 19 Nov 2026 06:27:03 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@stdevilgunjan/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[CoVariance]]></title>
            <link>https://medium.com/@stdevilgunjan/covariance-93b4f034d805?source=rss-cca56ab38963------2</link>
            <guid isPermaLink="false">https://medium.com/p/93b4f034d805</guid>
            <category><![CDATA[covariance]]></category>
            <category><![CDATA[correlation]]></category>
            <category><![CDATA[linear-regression]]></category>
            <dc:creator><![CDATA[Gunjan Grunge]]></dc:creator>
            <pubDate>Sat, 08 Jul 2023 15:00:21 GMT</pubDate>
            <atom:updated>2023-07-08T15:00:21.027Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*uhvbxn7_hTjRmklytI-H3Q.jpeg" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*mDVsDTf08lnm_RjRrrchaw.jpeg" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*4C40YDDpnSOX0INwd-HhmQ.jpeg" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*_MsdtGwqQvHfIWvkTERfmQ.jpeg" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*H-zf-U7GDph4QkGmsn7nsA.jpeg" /></figure><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=93b4f034d805" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Hypothesis Testing PART 2]]></title>
            <link>https://medium.com/@stdevilgunjan/hypothesis-part2-fb2b8bef87e3?source=rss-cca56ab38963------2</link>
            <guid isPermaLink="false">https://medium.com/p/fb2b8bef87e3</guid>
            <category><![CDATA[hypothesis]]></category>
            <category><![CDATA[type-1-error]]></category>
            <category><![CDATA[hypothesis-testing]]></category>
            <category><![CDATA[machine-learning]]></category>
            <category><![CDATA[statistics]]></category>
            <dc:creator><![CDATA[Gunjan Grunge]]></dc:creator>
            <pubDate>Mon, 19 Jun 2023 13:06:57 GMT</pubDate>
            <atom:updated>2023-06-20T06:54:15.965Z</atom:updated>
            <content:encoded><![CDATA[<p>A hypothesis is an assumption made without looking into the data, it is a statement that is made without even looking into how the data is distributed, avg etc.</p><p>There are two type of hypothesis that we make</p><ol><li>Null Hypothesis</li><li>Alternate Hypothesis</li></ol><p>Here, the alternate hypothesis is what matters as under this we are about to assume what is to be true or what we want to prove, while the null hypothesis is just the opposite and is a status quo where rejection leads to desired conclusion.</p><p>So let us talk with an example.</p><p>Lets say: We are looking into a packet of chocolate where the net weight mentioned is 500gm. Since it is labeled as 500gm we believe it to be true.</p><p>So the hypothesis can be from 2 different points of view, one from the manufacturer to ensure that it is not more than 500 gm and it is equal to 500gm while from a consumer point of view, we would want to prove that is either equal to 500gm but not less than 500gm. But we doubt the manufacturer and we come up with the hypothesis that the net weight of the packet is less than 500 gm this is what we want to prove and in absence of any evidence we will have to stand with the null hypothesis that each packet sold by the manufacturer equals to 500gm. Hence, we say we <strong>failed to reject</strong> the null hypothesis.</p><p>So, now what we do is collect x number of packets lets say 100 packets of chocolates all over the country and we have a random sample.</p><p>We measure the packets to find the mean of the packets</p><p>We then use the mean to test the null hypothesis stating that the mean should be equal to 500gm.</p><p>We also need to look into the error which is widely known as Type1 error and Type 2 error.</p><p>If a patient goes to a doctor with a certain disease and the doctor assumes it to be true then it is the correct decision. This is where the alternate hypothesis would be that the patient is not suffering from any disease. In this case we would fail to reject the null hypothesis while the patient is actually suffering.</p><p>Now if the patient was fine and it was just a mild symptom and not affected then and the doctor says that he is not suffering from any disease then this time the alternate hypothesis will win and we will reject the null hypothesis.</p><p>This is because the doctor would have assumed that he was affected and had made the assumption that he was suffering from a disease, however after diagnosis which will be the collection of evidence he had to reject the null hypothesis.</p><p>In the first case, the patient was suffering from a disease so let’s label the null hypothesis as <strong>TRUE </strong>and when the patient was just having mild symptoms then we label it as <strong>FALSE .</strong></p><p>I am just labeling this to come to a point, later it will be much more clear.</p><p>So coming back to the doctor and patient if in the first case the doctor assumed the patient to be suffering from a disease and after diagnosis he feels that the was not suffering from any disease and sends the patient without any medication while the in fact the patient was suffering then in this case we have made the <strong>type 1 error</strong>,it is also denoted by the letter <strong>alpha, </strong>And the second case he was not ill but the doctor said he was ill and gave him medication which would be an overdose or may have side effects this is called the <strong>type 2 error </strong>while this is denoted by <strong>beta.</strong></p><p>Now why alpha and beta?</p><p>Alpha is the confidence level of the test and so is the beta for type 2 error. Confidence level = 1 — alpha.</p><p>Which means you do not want to send home an ill patient saying nothing has happened to you. Hence the problem here we would face is in order to control one parameter we may blow up the other parameter, hence we have to balance it out.</p><p>The type 2 error is where we want to treat patients who are suffering rather than sending them home without medicine or giving medicine to a normal guy who is not suffering from any kind of illness. So in this case we maximize the power for fixed alpha and we keep the alpha low.</p><p>I know this explanation may not have made sense to you, but gradually it will.</p><p>The test also depends on the data type we are conducting the test on, having said that we can differentiate the data type in two broad categories which would be</p><ol><li>Continuous : For continuous data type, we would want to know are different samples coming from the same population and are close to the mean of population, this is where ANOVA comes into the picture.</li><li>Categorical : These are mostly a <strong>yes and no</strong> kind of data set. Here we would be inclined towards asking the questions like are the proportion of Yes in this group equal to the proportion of yes compared to the other group, if not equal then what is the difference. And this leads to <strong>2 proportion test</strong></li></ol><p><strong>So , how do we execute a test?</strong></p><p>1st we should have a clear argument or we can say a target what we want to prove or what questions are we asking</p><p>2nd state an alternative hypothesis, with the counterpart being the null hypothesis</p><p>3rd Determine the confidence level which the alpha value. Type1 error rate as alpha and which is widely assumed to be 0.05 unless you have a different number or value given to you.</p><p>4th State the decision rule, when to reject/fail to reject null hypothesis.</p><p>5th Gather the sample data</p><p>6th Calculate the test</p><p>7th State your statistical conclusions based on a <strong>P-value</strong></p><p>8th Make inference based on the conclusion.</p><p><strong>So we need the P value to make the decision. This value will tell us what is the confidence that we have in the null hypothesis.</strong></p><p>If the P value is low then we do not have much confidence in the null hypothesis and we would reject the null hypothesis while a higher p value will lead to failure to reject the null hypothesis.</p><p>So if</p><p>P &lt; alpha we reject</p><p>And if p &gt; alpha we fail to reject</p><p>As we know by now we have various types of hypothesis tests depending on the type of data we have. We can list them as</p><ol><li>Single sample or two or more sample</li><li>One tailed or two tailed</li><li>Test of mean,proportion of variance.</li></ol><p>Going back to the packet to chocolates we were concerned about, if we say that the packet of chocolates mean volume is not equal to 500 gms then we our</p><p>alternate hypothesis is mean is not equal to 500gm</p><p>Null hypothesis would be that mean = 500gm</p><p>A one-tailed test looks for an “increase” or “decrease” in the parameter</p><p>whereas a two-tailed test looks for a “change” (could be increase or decrease) in the parameter.</p><p>Therefore, if we see words such as “increased, greater, larger, improved and so on”, or “decreased, less, smaller and so on” in the original claim of a question (&gt;,&lt;are used in H1), a</p><p>a one-tail test is applied.</p><p>If words such as “change, the same, different/difference and so on” are</p><p>used in the claim of the question (≠is used in H1), a two-tailed test is applied</p><p>In case of our chocolates we are making a statement that mean = 500gm</p><p>Mean 500gm</p><p>This would be a Two Tailed test while if we had claimed that the packet has <strong>less than</strong> 500 gm then it would have been a one tailed test.</p><p>The purpose or the motivation behind conducting this test is to get a value to compare if this value is large or small.</p><p>Having said that, the mean may not be equal to 500 even if we draw enough samples as there may be a variation by certain milligram, now the question arises how much variance is acceptable to fail to reject the null hypothesis or reject the same ?</p><p>Now, lets measure the avg in a standardized way:</p><p>So my difference would be</p><p>Sample Mean — 500 = X</p><p>Now how do we determine if X is large or small or if it is adequate. Here, we let the nature of statistics decide, and the nature to determine what is less or more would be ? : comment your answer before you move on</p><p>In order to determine what is large or a small variation we would need figure out the standard way of variation we move on to the standard error to do this job this is where we go back to</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/179/1*b5uNtr8wNVX7XD-97zAiqQ.png" /></figure><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=fb2b8bef87e3" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Hypothesis Testing PART 1]]></title>
            <link>https://medium.com/@stdevilgunjan/hypothesis-testing-part-1-4d54c59654a8?source=rss-cca56ab38963------2</link>
            <guid isPermaLink="false">https://medium.com/p/4d54c59654a8</guid>
            <category><![CDATA[data-science]]></category>
            <category><![CDATA[statistical-analysis]]></category>
            <category><![CDATA[hypothesis-testing]]></category>
            <category><![CDATA[statistics]]></category>
            <category><![CDATA[data-visualization]]></category>
            <dc:creator><![CDATA[Gunjan Grunge]]></dc:creator>
            <pubDate>Tue, 13 Jun 2023 17:17:52 GMT</pubDate>
            <atom:updated>2023-06-13T17:25:18.581Z</atom:updated>
            <content:encoded><![CDATA[<p>This is technique in Inferential statistics, where we analyze the data and inference statistics with the purpose of Inferring information from it, while descriptive statistics which identifies the data describe it and inform the patterns etc.</p><p>Concepts of Sampling distribution:</p><p>Why do we need sampling?</p><p>Sampling is required to represent the data, which the popluation and while sampling we are actually interested in the population and the sample itself.</p><p>So we calculate the mean of the median of the population to relate the corresponding sample.</p><p>Hence, sampling distribution which is the distribution of a particular sample statistics of all possible sample that can be drawn from a population</p><p>So, from the population we can take a sample, and calculate any statistics which can later be used as a parameter to infer any information from the population. This is where we need to understand how we can represent or connect the sample with the population so that we can make the inference.</p><p>And here we come to “Sampling distribution” : this is where we take an assumption that all the samples which could have been observed have been observed. Since we only have 1 set of samples the process here would be to check the population via the parameter and the statistic and draw the sample.</p><p>Since the population of the data is random and may not be the same as usual, it may be hard to digest the fact that the same math can help us to drive the sample. Hence, we make certain assumptions and also have few concepts which we can follow to achieve the same.</p><p>The first one is <strong>Central Limit Theorem</strong>:</p><p>Under this concept we get to assume that when the data is sufficiently large, it follows a pattern. Eg: We have a population where mean= µ and the standard deviation =ó</p><p>These mean and Standard Deviation are the parameters that we talked about in the above paragraph.</p><p>Hence, the sampling distribution follows a normal distribution, also the sample will have its own mean and a standard deviation of ó ÷ √n, the n here denotes the sample size and is our Standard deviation of the population.</p><p>ó ÷ √n is also called a Standard Error. Hence, it is worth noting that higher the sample, i.e sample size close the population will be nearing to the mean of the population,which also means that if the sample mean = x̄ will almost be identical to µ</p><p>This opens up the option to make an assumption that “ if the population has normal distribution then even for a small sample size whatever has been described is correct. If the population is not normally distributed then we can hold on to same parameter and the calculation, provided we have another calculation which is the Z score or we can also call it z score transformation, .i.e</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/201/1*NL4IF64Z2c7zzYq5Z0sOkw.png" /></figure><p>This is the standardization of distribution.</p><p>Now, the Central Limit Theorem says, if the n is large then Z is appox normally distributed.</p><p>Hence, “Sampling <strong>Distribution of the mean</strong> of any independent random variable <strong>will be norma</strong>l “</p><p>Listing down the assumptions:</p><ul><li>The data must to randomly sampled</li><li>The sample value must be independent.</li><li>When the sample is drawn without replacement the sample size n should be not more than 10% of the population.</li></ul><p>Let’s prove this with an experiment:</p><p>Assuming a dice is rolled in sets of 4 trials and the faces are recorded. This is repeated for 30 days.</p><p>I made this table on excel using =RANDBETWEEN(1,6)</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/384/0*7or9mHSVgKMXA4Uz" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/382/1*olxZxMWpVHs-xXtY4e2O1A.png" /></figure><p>We can see that each day we get different mean and what we can do is plot a frequency table</p><p>First I took the unique values of mean and checked for its frequency :</p><p>Unique Mean</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/252/1*RX5cvvmKoPaaNE00I9fa4w.png" /></figure><p>When I plot this, I can see the following chart:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/478/1*ymqR2eNHiw3NnDn_Jz9dbw.png" /></figure><p>Which more or less is normally distributed as we can observe that the avg is tending towards the 3.25 mark as the mean for the population is 3.308 and we can see that the frequency also tends towards 3.25 and 3.50.</p><p>I will explain about the Hypothesis Testing in next part as knowing about Hypothesis Testing would need a background of these theories.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=4d54c59654a8" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Probability — Meaning and concepts]]></title>
            <link>https://medium.com/@stdevilgunjan/probability-meaning-and-concepts-28c0538675a8?source=rss-cca56ab38963------2</link>
            <guid isPermaLink="false">https://medium.com/p/28c0538675a8</guid>
            <category><![CDATA[probability]]></category>
            <dc:creator><![CDATA[Gunjan Grunge]]></dc:creator>
            <pubDate>Fri, 09 Jun 2023 13:42:35 GMT</pubDate>
            <atom:updated>2023-06-09T13:42:35.632Z</atom:updated>
            <content:encoded><![CDATA[<p><strong>Probability — Meaning and concepts</strong></p><p>Probability refers to chance or likelihood of a particular event taking place</p><p>An event is an outcome of an experiment</p><p>An experiment is a process that is performed to understand and observe possible outcomes.</p><p>Set of all outcomes of an experiment is called the sample space.</p><p>Probability of an event can always be 0–1 a pure number. We can denote these in percentages as well, Where 0 means the event is not likely to occur and 0.5 is 50% chance for the same to occur and 1 is 100% chance that the event will occur.</p><p><strong>Mutually Exclusive Events</strong> are those events where there is nothing common between them, specifically if <strong>ONE HAPPENS WE KNOW THAT THE OTHER WON’T HAPPEN</strong></p><p>Drawing 1 card which will be a Face card is an example of Mutually exclusive event, as When we draw one card, it can either be a king or a queen it is possible that it is not both the card, and if kind is picked it will surely not be queen on the card that is picked.</p><p><strong>Independent Events. : </strong>Two Events A and B are said to be independent if the occurrence of A is in no way influenced by the occurrence of B. Likewise occurrence of B is in no way influenced by the occurrence of A</p><p>This is different from Mutually Exclusive events, they are not Independent events, as occurrence of one tells us that the other will not occur, while in independent event, it tells nothing about the other occurrence and is only concerned with the current event,</p><p>Rules for Computing Probability :</p><p>If the events are mutually independent , which means if the event is like it cannot happen together then what is the chance of either one occurring? This is where we would apply the <strong>Addition Rule.</strong></p><p><strong>p(AuB) = p(A) + p(B)</strong></p><p><strong>Example</strong></p><p><strong>From a pack of well-shuffled cards, a card is picked up at random.</strong></p><ol><li><strong>What is the probability that the selected card is king or a queen</strong></li></ol><p>Let A = getting a King</p><p>Let B = getting a Queen</p><p>A = 4/52</p><p>B = 4/52</p><p>A+b = 8/52 = 2/13</p><p>Multiplication Rule:</p><p>The rule says when two events A and B are independent, the probability of the simultaneous occurrence of A and B equals the product of the probability A and the probability of B.</p><p>p(A n B)</p><p>Example: If the probability of getting A grade in X subject is 0.7 and the probability of getting A grade in Y subject is 0.5 and these are two independent subjects then the probability of getting a grade in both the subjects are 0.7 * 0.5 = 0.35 that is also a 35% chance.</p><p><strong>What happens when the events are not independent ?</strong></p><p>The rule says that the probability of the intersection of the event a and b equal to the product of A and B equal to the product of the probability of A and B given that A has happened or known to you. p(A|B) or p(B|A) is called conditional probability of B given the fact that A has happened.</p><p>Eg:</p><p>p(AnB) = p(A) X p(B|A)</p><p>i.e Probability of A and B happening = probability of A happening X</p><p>probability of B given A has</p><p>happened.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=28c0538675a8" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Coefficient of Variation]]></title>
            <link>https://medium.com/@stdevilgunjan/coefficient-of-variation-e19c3a7f5a82?source=rss-cca56ab38963------2</link>
            <guid isPermaLink="false">https://medium.com/p/e19c3a7f5a82</guid>
            <category><![CDATA[box-plot]]></category>
            <category><![CDATA[statistics]]></category>
            <dc:creator><![CDATA[Gunjan Grunge]]></dc:creator>
            <pubDate>Mon, 05 Jun 2023 18:21:42 GMT</pubDate>
            <atom:updated>2023-06-05T18:21:42.673Z</atom:updated>
            <content:encoded><![CDATA[<p><strong>Coefficient of Variation</strong></p><p>Let us take an example to compare two Salesman,</p><p>Where SalesMan A sells cars with an average of 50 units. The Standard Deviation is 5 units, while SalesMan B sells 75 units with Standard Deviation of 25 units.</p><p>The word we need to focus here is <strong>Deviation, </strong>divergence,difference,disparity. Meaning that</p><p>The salesman A may sell 5 units more or less.Whiles Sales B may sell 25 units more or less.</p><p>Depending on risk and since we have taken an example of a car dealer, one would not risk to depend on a salesman who may sell 25 units more or even less. While the variation or we can call it the divergence on the other salesman is only 5 units.</p><p>Although, SalesMan B has given me more sales this year, I would not be able to rely on him for next year due to the variance of 25 units, which he can sell more or less. However, I will surely bank on the salesman who only sold 50 units with the difference of only 5 units.</p><p>Now we go into Coefficient of Variation : We calculate CV as</p><p>StandardDeviation/Mean</p><p>So for salesman A it would be 5/50 = 0.10 which is 10% and for Salesman B it would be 25/75 = 0.33 i.e 33%</p><p>Hence, it is always advisable to determine and iron out risk by ignoring the average and looking into the variance and dispersion of the data.</p><p><strong>The Empirical Rule</strong></p><p>The empirical rule approximates the variation of data in bell shaped distribution. Approximately 60% of the data is a bell shaped distribution within 1 standard deviation of the mean.</p><p>So we determine here that mean +- 1 standard deviation covers 68% of the data.While,</p><p>mean +- 2 standard deviation covers 95% of the data and</p><p>mean +- 3 standard deviation covers 99.7% of the data</p><p><strong>The Chebyshev Rule</strong></p><p>The rule states regardless of how the data are distributed at least(1 -1/k2) x 100% of the values will fall within K standard deviation of the mean (for k &gt;1)</p><p>We use the Chebyshev rule when the data is not bell shaped.</p><p>Eg: When k = 2</p><p>It is 1–½*2</p><p>Or 1–¼</p><p>Or 1–0.25</p><p>75%</p><p>So 75% of the data will be within mean +- 2 standard deviation, when the data is not normally distributed.</p><p><strong>The BOX PLOT</strong></p><p>The five number that helps describe the center, spread and the shape of the data are</p><ol><li>X smallest</li><li>First Quartile( Q1)</li><li>Median(Q2)</li><li>Third Quartile (Q3)</li><li>X Largest</li></ol><figure><img alt="" src="https://cdn-images-1.medium.com/max/606/1*6LoV1hO_TxWzKF3HJtAsXg.png" /></figure><ul><li>The Box refers to Q1 Q2 and Q3</li><li>The Line through the box is median</li><li>The Box Starts from Smallest and ends at the largest Value</li><li>The Box shows the central tendency of the data</li><li>The dispersion of the data</li><li>The skewness of the data</li><li>The outliers in the data</li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/707/1*-vuwVH04Z2RSpM8YWwKyuA.png" /></figure><p>The box will be shrinked towards the left and a long line can be seen towards the right, this is called a right skewed data</p><p>The Line in red extends up to 1.5 times the IQR and stops. And if there are values in the dataset beyond that number it will represent that with a dot and throw it outside the box, which we term as outlier.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/640/1*mLZc2WnBhigCu9Vq2H2syg.png" /></figure><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=e19c3a7f5a82" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Statistical Methods]]></title>
            <link>https://medium.com/@stdevilgunjan/statistical-methods-168eb6bd38df?source=rss-cca56ab38963------2</link>
            <guid isPermaLink="false">https://medium.com/p/168eb6bd38df</guid>
            <category><![CDATA[standard-deviation]]></category>
            <category><![CDATA[statistics]]></category>
            <dc:creator><![CDATA[Gunjan Grunge]]></dc:creator>
            <pubDate>Sat, 03 Jun 2023 17:46:43 GMT</pubDate>
            <atom:updated>2023-06-03T17:46:43.806Z</atom:updated>
            <content:encoded><![CDATA[<p><strong>Classification:</strong></p><p>Classification techniques help in segmenting the customers into appropriate groups based on key characteristics. Eg: Segmenting the customer in “Long Term Customers, Medium Term Customer or Brand Switchers” or even make a classification of customers as buyer and non buyers.</p><p><strong>How would this help ?</strong></p><p>This would help to understand the market behavior and will provide a better picture of what to sell and when to sell and even further whom to sell. Which will help a brand to adopt necessary strategies to reach their Goal.</p><p><strong>Pattern Recognition:</strong></p><p>A distribution of the data helps us understand how the data is distributed and how if there is a pattern, this is where we can take help of visualization such as Histogram, Box Plot and Scatter plot as well. We can name N number of visualizations to seek a pattern in the data. Hence, this is descriptive statistics; which helps us summarize the data and represent the data in a visual form to extract patterns and behavior of the set. This will also help us extract the actual information and elements noise out of the data, of course we would need to make a few calculations and implement a few techniques to decipher the dataset as per the requirements.</p><p><strong>Association: </strong>The Relationship between variables, objects. Hence, these relationships between variables are a result of calculations and assumptions. For an example we can say,x number of customers who buy milk will also buy bread. Which will now take us to the probability classes, therefore we can come up with the statement that if x unit of milk packets are sold then there is X% probability that x no of breads will also be sold along with it. <strong>How would this help?: </strong>This will help the brand plan out discount offers, plan the storage and even align items in a store inorder to maximize sales.</p><p><strong>Predictive Modeling</strong>: Models like regression is a good example of predictive modeling where we make computations based on the formula</p><p><em>“y= a+bx+e’’ </em>here Y is the DEPENDENT variable X is the INDEPENDENT variable, a is the intercept, b is the slope, e is the residual . This is where we give an input variable as X and it will be a prediction on Y .</p><p>We can have such predictive modeling used in more advanced ways where we can predict a target variable as well as classify using Logistic Regression and Artificial Neural Network.</p><p>Comparing Descriptive Statistics and Inferential Statistics we can say the Descriptive Statistics Is associated with the Data Summarization, Graphs/charts and tables, which talks about the data while the later talks about the method use to talk about a population parameter from a sample, while Descriptive Statistics talks about what the data is all about where the later talks about what could have been.</p><p>Descriptive Statistics gives us a picture of what has already happened while Inferential Statistics gives us an idea of what can happen.</p><p><strong>Dispersion:</strong></p><p>Measure of dispersion indicates how large the spread of the distribution is around the central tendency.</p><p>Range: It is the simplest of all measures of dispersion. It is calculated as the difference between max and min value in the dataset, the range tells us how the data is spread. However, if the max or min value is extreme then the range should not be used, to overcome this problem we move towards IQR (inter quartile range)</p><p>IQR is the range computed on the middle 50% of the observations after eliminating the highest and lowest 25% of the observations in a data set that is arranged in ascending order. IQR is less affected by outliers.</p><p>IQR = Q3 — Q1.</p><p>Q1 is 25% which is the min and is called the lower quartile and the max 25% is called the upper quartile. We ignore both upper and lower quartile and consider the range which is within the 50% which is the inter quartile range.</p><p>Eg: Data = 9,10.5,11,11,12,12,14,14,18</p><p>So the highest value is 18 and lowest is 9 so we will ignore these values and calculate the iqr within the range (10.5,11,11,12,12,14,14) and now the max is 14 and min is 10.5 so :</p><p>Lower quartile which is the 25% falls between 10.5 and 11 so the avg of 10.5 and 11 is 10.75</p><p>Upper quartile which is the other 25% falls between 14 and 14 hence, the avg of 14 and 14 so its 14 hence,</p><p><strong>IQR = 14–10.75 = 3.25</strong></p><p>In this case, it means that the middle 50% of the data and within the range of approximately 10.5 to 14. Therefore, the IQR helps in understanding the spread or variability of the data, rather than indicating a specific assured value.</p><p>if the IQR is 3.5, it means that the middle 50% of the returns fall within a range of approximately 10.5 to 14. This tells us that most of the values are clustered within this range.</p><p>That brings the discussion to <strong>Standard Deviation</strong>: in order to understand Standard Deviation, we also need to get used to the term called “Variance”. Technically Standard Deviation is the Square Root of Variance.</p><p>Let us take example of this data</p><p><strong>Data = 9,10.5,11,11,12,12,14,14,18</strong></p><p>The first step is to calculate the MEAN, th<strong>e arithmetic mean :</strong></p><p>If we calculate the avg</p><p>Data = 9,10.5,11,11,12,12,14,14,18</p><p>def Average(lst):</p><p>print(sum(lst) / len(lst))</p><p>Average(Data)</p><p>It will print 12.39 that is our MEAN.</p><p>The Second Step is to calculate the difference between the value and the mean.</p><p>So, it will start as 9–12.39 = — 2.39</p><p>Data = [9, 10.5, 11, 11, 12, 12, 14, 14, 18]</p><p># Calculate the average</p><p>avg = sum(Data) / len(Data)</p><p># Subtract the average from each element in Data</p><p>subtracted_data = [x — avg for x in Data]</p><p>print(subtracted_data)</p><p>[-2.33, -0.83, -0.33, -0.33, 0.67, 0.67, 2.7, 2.67, 6.67]</p><p>The Third step is to square the subtracted_data:</p><p>squared_data = [x ** 2 for x in subtracted_data]</p><p>print(squared_data)</p><p>[11.48, 3.57, 1.93, 1.93, 0.15, 0.15, 2.6, 2.6, 31.48]</p><p>Finally we calculate the Total Sum of Square:</p><p>sum_of_rounded_data = sum(rounded_data)</p><p>print(sum_of_rounded_data)</p><p>55.89</p><p>Now We find the avg of the rounded data:</p><p>avg_of_rounded_data = sum(rounded_data) / (len(rounded_data)-1)</p><p>print(round(avg_of_rounded_data,2))</p><p>6.99</p><p>The Variance is. 6.99</p><p>Remember we took the (no of observations — 1) hare, and not the regular avg.</p><p>The standard deviation is the square root of 6.99</p><p>import math</p><p>sqrt_of_avg = math.sqrt(avg_of_rounded_data)</p><p>print(round(sqrt_of_avg,2))</p><p>The Standard Deviation is 2.64.</p><p>Why did we use Square root?</p><p>As the avg_of_rounded_data was the avg of squares that we had taken, hence square root to nullify the same.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=168eb6bd38df" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[<html> Basic </html>]]></title>
            <link>https://medium.com/@stdevilgunjan/html-basic-html-1c178688bb28?source=rss-cca56ab38963------2</link>
            <guid isPermaLink="false">https://medium.com/p/1c178688bb28</guid>
            <category><![CDATA[block-element-modifier]]></category>
            <category><![CDATA[html]]></category>
            <category><![CDATA[elements]]></category>
            <dc:creator><![CDATA[Gunjan Grunge]]></dc:creator>
            <pubDate>Tue, 27 Dec 2022 12:16:20 GMT</pubDate>
            <atom:updated>2022-12-27T12:16:20.545Z</atom:updated>
            <content:encoded><![CDATA[<p>!</p><p>We start HTML with “!” these days, thanks to emmit. By just one key stroke we get the boilerplate</p><p>A boilerplate looks like this:</p><pre>&lt;!DOCTYPE html&gt;<br>&lt;html lang=&quot;en&quot;&gt;<br>&lt;head&gt;<br> &lt;meta charset=&quot;UTF-8&quot;&gt;<br> &lt;meta http-equiv=&quot;X-UA-Compatible&quot; content=&quot;IE=edge&quot;&gt;<br> &lt;meta name=&quot;viewport&quot; content=&quot;width=device-width, initial-scale=1.0&quot;&gt;<br> &lt;title&gt;Document&lt;/title&gt;<br>&lt;/head&gt;<br>&lt;body&gt;<br> <br>&lt;/body&gt;<br>&lt;/html&gt;</pre><p>We are ready to start writing HTML.</p><p>Before we beign we should note that “<strong>HTML is the standard markup language for documents designed to be displayed in a web <em>browser</em></strong>”.</p><p>Hence, it is not a programming language and therefore even if you end up writing a wrong code it would not give you an error.</p><p>if you ask me: <strong>ETML5</strong></p><p>I would say: HTML is a Language made up of elements which can be applied to pieces of text to give them different meaning in a document.</p><p>What is an element? you may ask:</p><p>Well, an element is a part of a webpage. An element may contain a data or a group of images or nothing.</p><p>eg:</p><p>&lt;p&gt; is a tag</p><p>&lt;p class=”anyname”&gt; this is a class name we call it a value of the attribute, class is an attribute and “anyname” is a value.</p><p>The below image is an Element. Which contains Opening tag, closing tag and also Attributes along with its corresponding value for the content and ending with a closing tag.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/522/1*s4gkVGt0voxJvy7jtnwDBw.png" /><figcaption>ELEMENT IN HTML</figcaption></figure><p>Understanding this is very important as it will help you later in the long run to research and also write your code logic<strong>.</strong></p><p><strong>Nesting Element: </strong>Eelement can be placed within other element, this is called nesting.</p><h3><em>&lt;p&gt; My dog is &lt;b&gt; very &lt;b&gt; cute &lt;/p&gt;</em></h3><p>Here, we nested &lt;b&gt; element inside &lt;p&gt;</p><p><strong>BLOCK and InLine Elements</strong></p><p>Block Elements are like blocks it will take a lot of space, it would not allow any another element to take its place, even if you just write a single letter it will take the entire space the entire row while inline would take the space that it requires to fit.</p><p><em>Remember</em>: A block element would not be nested inside an inline element, But can be nested inside another block element.</p><p>Examples of a block level elements are:</p><ol><li>Address</li><li>Article</li><li>aside</li><li>blockquotes</li><li>details</li><li>dialouge</li><li>div</li><li>figcaption</li><li>footer</li><li>form</li><li>h1</li><li>h2</li><li>header</li><li>main</li><li>nav</li><li>section</li><li>li</li><li>ol</li><li>main</li><li>p</li><li>hgroup</li></ol><p>and the list can go on.</p><p>while for inline elements we have</p><ol><li>em</li><li>a</li><li>b</li><li>strong</li></ol><p>very few which we normally do not use much and use CSS to help us with these.</p><p>The point to focus is mostly on &lt;img&gt; and &lt;usemap&gt; tags, these are magical tags which will help you make your images and clips stand out and interactive.</p><p>usemap uses a concept where you can select an area of an image or a clip and you can make it interactive where if the user clicks on it then it can redirect it to the actual source of where ever you want them to go.</p><p>Usemap is a complimantary arttribute which goes with the &lt;img&gt; tag.</p><p>In order to understand this better referring the MDN documention is recommended.</p><p>while I can give an example here if you can relate :</p><pre>&lt;map name=&quot;myareamap:&gt;<br>&lt;area shape=&quot;rec&quot; coords=&quot;89 34 98 33&quot; href=&quot;www.theweb.com&quot; alt=&quot;theweb&quot;&gt;<br>&lt;/map&gt;<br><br><br>&lt;img usemap=#myareamap <br>src =&quot;thesource&quot;<br>alt=&quot;altnameofthesource&quot;&gt;<br></pre><p>The flow can be this :</p><ol><li>Select an image that you want to render</li><li>You will need to identify the coordinates on that image which needs to be hyperlinked ( as in where and when if clicked it will take you to a different web/portal)</li><li>By using &lt;map&gt; you will create a shape which can be anything like circle, rectangle and give it a coordinate to represent the shape on that image.</li><li>finally use the usemap attritbute with the vlaue in the image tag and you are all set.</li></ol><p>NICE AND EASY.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=1c178688bb28" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[From C.A Dropout To MLops and Front End Developer]]></title>
            <link>https://medium.com/@stdevilgunjan/from-c-a-dropout-to-mlops-and-front-end-developer-25bdc78bf72e?source=rss-cca56ab38963------2</link>
            <guid isPermaLink="false">https://medium.com/p/25bdc78bf72e</guid>
            <category><![CDATA[change-your-life]]></category>
            <category><![CDATA[developer-tools]]></category>
            <category><![CDATA[front-end-developer]]></category>
            <category><![CDATA[machine-learning]]></category>
            <category><![CDATA[journey]]></category>
            <dc:creator><![CDATA[Gunjan Grunge]]></dc:creator>
            <pubDate>Mon, 22 Aug 2022 16:47:08 GMT</pubDate>
            <atom:updated>2022-08-22T16:47:08.218Z</atom:updated>
            <content:encoded><![CDATA[<p>I started off with being a Chartared Accountant, predicting something with the magic of Maths was something I always enjoyed. However, social resposibilities deffered the idea and I had to stop my Journey.</p><p>After, few years I realised about Analytics and Predictive modeling, I started working with a Team of Analyst in an org, got to know that a combination of Data and maths added to it, can make wonders, that is when I over heard about A.i, facinating stuff at that time. As usual I thought its all about talking robots and nothing more than it.</p><p>Few years down the line, I got into an org that used A.i and I had to do a small part of Data Exploration for the team as by now, I was a Pro Analyst.</p><p>Now, A.i started making sense to me, I dont know what do you refer to the year 2020 as, but with all the darkness around, the one ray of light that I saw was taking up a course onling for AiML as going out was restricted. Cut to shot, I now work as an Mlops… but that did not end there.</p><p>I was so facinated with Ai and Machine learning that I wanted to share my models and ideas to the world, so that may be someone would benefit from model I have build for them.</p><p>Now, I wanted to design a UI or webapp which would make my models accessable. I know you are thinking “dude share it in github”, but that is for pros, there are 1000s of pros on github who may have build millions of models and are still building as I type and while you read, but that does not help a beginner, or a semi pro. Or a statup or may be an enterprenure, who just wants to test and does not want to hire someone just to do this may be, or may be someone may need to finish the model within a dealine, end no of scenarious, my idea here is I just want my models to be accessable.</p><p>we can debate on this sometime later, while getting back to my front end journey, I started off with tkinter in python, I was like naaa. I want a full fucntional website. I checked fiverr, but wasnt happy as I would not be able to keep adding stuff, I need to hire someone, so I hired myself, got ready for another job at the same time which would not violet any contract. So, to get job ready, I started learning HTML, CSS and few days I was able to design a sweet design for a friend who offered discount on Airline tickets over instagram and whatsapp, well now he has a website build by me with my Trademark on it, oh yes I filed for a trademark.</p><p>Be it Ai or Front end designing,I give my 100% to whatever I do. While I was getting good with HTML and CSS, I also knew, javascript is an essential to build a good web application, so I resreached,went through few Youtube content, but the best was <a href="https://scrimba.com/learn/learnjavascript">https://scrimba.com/learn/learnjavascript</a> <a href="https://medium.com/u/30d952e8c7e6">Per Harald Borgen</a>.</p><p>Who also motivated to write this article on Medium, while I am confident about my JScript jouerney. I would like to thank the entire Scrimba community and faculty for encouraging and teaching the best way possible.</p><p>Well if this article is not to old *for future* you may want to check my web TheAiChamp.com, as on 22/08/2022 it reads coming soon, oh yeah, dont miss the tagline on the logo.</p><p>//I typed it all at one go. please excuse me for any spelling or grammatical errors.</p><p>Thanks,</p><p>console.log(“$(Gunjan)!”)</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=25bdc78bf72e" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>