<?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 Ishani Udeshika on Medium]]></title>
        <description><![CDATA[Stories by Ishani Udeshika on Medium]]></description>
        <link>https://medium.com/@Ishani.Udeshika?source=rss-ead4128890d1------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*bHdxiQlsyLRAvQ6KOfw_cQ.jpeg</url>
            <title>Stories by Ishani Udeshika on Medium</title>
            <link>https://medium.com/@Ishani.Udeshika?source=rss-ead4128890d1------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Fri, 20 Nov 2026 07:45:20 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@Ishani.Udeshika/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[Leveraging Monte Carlo Simulation in AI for Predictive Analytics]]></title>
            <link>https://medium.com/@Ishani.Udeshika/leveraging-monte-carlo-simulation-in-ai-for-predictive-analytics-1d5fae22e79d?source=rss-ead4128890d1------2</link>
            <guid isPermaLink="false">https://medium.com/p/1d5fae22e79d</guid>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[statistical-analysis]]></category>
            <category><![CDATA[monte-carlo-simulation]]></category>
            <category><![CDATA[predictive-analytics]]></category>
            <category><![CDATA[real-time-streaming-data]]></category>
            <dc:creator><![CDATA[Ishani Udeshika]]></dc:creator>
            <pubDate>Thu, 03 Jul 2025 11:19:24 GMT</pubDate>
            <atom:updated>2025-07-03T11:21:07.602Z</atom:updated>
            <content:encoded><![CDATA[<h3>1. Introduction</h3><p>Artificial Intelligence (AI) has revolutionized nearly every industry, driving advancements in automation, decision-making, and predictive analytics. However, AI systems rely heavily on statistical modeling techniques to enhance their performance and reliability. One such technique, <strong>Monte Carlo Simulation (MCS)</strong>, plays a crucial role in handling uncertainty, improving AI-driven predictions, and optimizing decision-making processes.</p><p>While AI models excel at recognizing patterns and making predictions, they often struggle with uncertain or incomplete data. This is where Monte Carlo Simulation proves invaluable, providing probabilistic insights that complement AI models. By simulating thousands of possible outcomes, MCS enhances forecasting accuracy in dynamic environments, making it a powerful tool for risk assessment and decision-making.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*d_TIHeUNbqV_Eh_eT41YEw.png" /><figcaption>Figure 1: Machine Learning workflow</figcaption></figure><h3>2. Challenges in AI Predictions</h3><p>Despite its advantages, artificial intelligence comes with inherent challenges:</p><ul><li><strong>Data Dependency</strong>: High-quality, large datasets are required for effective model training.</li><li><strong>Overfitting &amp; Underfitting</strong>: Models may either memorize noise or fail to capture underlying patterns.</li><li><strong>Black Box Nature</strong>: Many AI models, particularly deep learning, lack interpretability.</li><li><strong>Data Drift</strong>: Evolving data distributions reduce model accuracy over time.</li></ul><p>Consider a scenario where a company wants to predict the sales of a newly launched product:</p><ol><li>Is there sufficient historical data to predict its demand?</li><li>How can uncertainty in customer preferences and market conditions be managed?</li></ol><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*139VZOJrnJcCMrRU" /><figcaption>Figure 2: Machine learning failures</figcaption></figure><p>When traditional AI models struggle with such scenarios, Monte Carlo Simulation provides a robust approach to handling uncertainty and enhancing predictive accuracy.</p><h3>3. Monte Carlo Simulation: A Statistical Approach to Uncertainty</h3><p>Monte Carlo Simulation (MCS) is a probabilistic modeling technique used to assess risk and uncertainty in prediction and decision-making. By relying on random sampling and statistical analysis, MCS generates a range of possible outcomes along with their probabilities, providing deeper insights into potential risks and opportunities.</p><h3>Why Use Monte Carlo Simulation?</h3><p>Unlike deterministic models, which provide a single-point estimate, Monte Carlo Simulation incorporates randomness into predictions. This makes it particularly useful when dealing with unpredictable environments. For example, predicting delivery time for a package involves various uncertainties such as traffic conditions, route diversions, and weather changes. MCS accounts for these variables by running multiple simulations to estimate probable outcomes.</p><h3>How Does Monte Carlo Simulation Work?</h3><p>The Monte Carlo method operates on the principles of probability distribution modeling:</p><ol><li><strong>Define Input Variables</strong>: Identify uncertain parameters (e.g., demand fluctuations, inventory levels, or financial market trends).</li><li><strong>Assign Probability Distributions</strong>: Represent future data points using Normal, Uniform, Exponential, or Log-normal distributions.</li><li><strong>Run Simulations</strong>: Generate thousands of random scenarios based on the assigned distributions.</li><li><strong>Analyze Results</strong>: Interpret the range of possible outcomes and their likelihoods.</li></ol><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*0Pdx1I4RqkBcShxx" /><figcaption>Figure 3: Steps of the Monte Carlo Simulation</figcaption></figure><p>Figure 3: Steps of the Monte Carlo Simulation</p><p>By performing repeated simulations, businesses can gain a probabilistic understanding of potential risks and make data-driven decisions.</p><h3>4. Use Cases of Monte Carlo Simulation</h3><p>Monte Carlo Simulation is widely used across industries to mitigate risks, optimize decision-making, and enhance predictive analytics. By running multiple simulations, businesses can evaluate potential scenarios and take proactive measures to improve outcomes. Below are some key use cases where Monte Carlo Simulation proves invaluable.</p><ul><li><strong>Finance and Investment: </strong>Monte Carlo Simulation helps financial institutions assess market risks, optimize portfolios, price derivatives, and estimate credit risk by simulating thousands of economic scenarios.</li><li><strong>Manufacturing and Supply Chain: </strong>Manufacturers and logistics companies use Monte Carlo Simulation to optimize production schedules, predict equipment failures, manage inventory levels, and assess supply chain disruptions.</li><li><strong>Engineering and Design: </strong>Engineers apply Monte Carlo Simulation to predict material fatigue, structural failure, and equipment reliability, ensuring safer designs and optimizing resource allocation in industries like construction and oil exploration.</li><li><strong>Healthcare and Medicine: </strong>In healthcare, Monte Carlo Simulation aids in disease spread modeling, clinical trials, radiation therapy planning, and hospital resource management to enhance patient outcomes and operational efficiency.</li><li><strong>Insurance and Actuarial Science: </strong>Insurance companies utilize Monte Carlo Simulation to model catastrophic risks, price policies accurately, and estimate life expectancy for sustainable underwriting and financial planning.</li><li><strong>Energy and Utilities: </strong>Monte Carlo Simulation supports energy companies in forecasting power demand, optimizing renewable energy storage, predicting oil price fluctuations, and improving power grid reliability.</li><li><strong>Telecommunications and Network Design: </strong>Telecom providers leverage Monte Carlo Simulation to optimize network traffic, predict data usage fluctuations, and assess cybersecurity risks for enhanced infrastructure resilience.</li></ul><h3>5. Monte Carlo Simulation in Enterprise AI: Enhancing Predictive Analytics</h3><p>In my work within enterprise Data Science and AI, I have integrated <strong>Monte Carlo Simulation</strong> into AI-driven analytics solutions to enhance predictive accuracy across diverse business domains. These simulations have played a vital role in enabling enterprise solutions to make confident, data-informed decisions under uncertainty.</p><ul><li><strong>IT Infrastructure &amp; Network Management</strong><br> A Quality of Service (QoS) solution was developed using a resource monitoring agent to oversee production systems across IT infrastructure and network layers. By applying Monte Carlo Simulation, was able to <strong>proactively forecast potential system failures</strong>, ensuring compliance with SLAs and minimizing downtime risks.</li><li><strong>Video Streaming Optimization</strong><br>Within a Cognitive Vision based solutions, Monte Carlo Simulation was used to <strong>simulate variable streaming conditions</strong>, enabling real-time <strong>bandwidth allocation and stream quality optimisation</strong> for AI-powered video analytics.</li><li><strong>Retail Demand Forecasting</strong><br> In retail analytics, Monte Carlo techniques were applied to <strong>model demand variability</strong>, allowing for more reliable inventory planning and helping to reduce both stockouts and surplus.</li></ul><p>By incorporating Monte Carlo Simulation into enterprise-scale AI and analytics frameworks, I’ve contributed to solutions that significantly improved <strong>forecast precision</strong>, <strong>risk quantification</strong>, and <strong>strategic decision-making</strong> for large-scale operations.</p><h3>6. Conclusion</h3><p>As industries continue to evolve in an increasingly data-driven world, handling uncertainty remains a critical challenge. While AI provides powerful predictive capabilities, it is not always sufficient when data is limited or when uncertainty is high. <strong>Monte Carlo Simulation bridges this gap</strong>, offering a structured approach to probabilistic modeling and risk analysis.</p><p><em>Are you ready to embrace the power of predictive analytics?</em> Let’s explore how Monte Carlo Simulation can transform your business operations!</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=1d5fae22e79d" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[My Experience with the Exam DP-100: Designing and Implementing a Data Science Solution on Azure]]></title>
            <link>https://medium.com/@Ishani.Udeshika/my-experience-with-the-exam-dp-100-designing-and-implementing-a-data-science-solution-on-azure-0f655fab49e2?source=rss-ead4128890d1------2</link>
            <guid isPermaLink="false">https://medium.com/p/0f655fab49e2</guid>
            <category><![CDATA[ml-pipeline]]></category>
            <category><![CDATA[mlops]]></category>
            <category><![CDATA[dp-100-exam]]></category>
            <category><![CDATA[azure-machine-learning]]></category>
            <category><![CDATA[machine-learning]]></category>
            <dc:creator><![CDATA[Ishani Udeshika]]></dc:creator>
            <pubDate>Wed, 27 Dec 2023 03:12:06 GMT</pubDate>
            <atom:updated>2023-12-27T03:17:38.615Z</atom:updated>
            <content:encoded><![CDATA[<p>We all often wonder about what it takes to build a successful data science project. It’s common to find discussions covering only a few steps like data collection, preprocessing, exploratory data analysis (EDA), and model training. But is this all there is to a data science project?</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/960/1*F-uX-qmfUK0bJzIQSIx_Ag.png" /></figure><p>If you do a reality check, there’s more to it, both before and after these steps. A critical part often overlooked is the continuous monitoring after post deployment.. The Data Science Projects aren’t just about immediate outcomes; they require monitoring throughout their lifetime in order to ensure the Machine Learning models serve its purpose.</p><p>Like many others, I initially focused on only a few stages of the data science workflow based on the academia experience. But when I started to preparing for the DP-100: Designing and Implementing a Data Science Solution on Azure exam, the initial perspective I had changed. It was a tough call at first, but then again I realized it was worth taking the risk. The exam wasn’t just about Azure ML portal navigation; it delved deeply into practical scenarios, presenting issues and demonstrating their resolution using Azure ML.</p><p>During this process, I discovered the significance of MLOps and ML pipelines. These learning helped me to understand the fact how easily these can be performed using Azure ML, a platform that simplifies the entire data science workflow.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*Ui3K9BT8C_oVwFU60I4w1Q.jpeg" /></figure><p>The beauty of the exam lies in its practical relevance. The concepts were not just about theories; Whether it was streamlining model deployments or optimizing workflows, each concept found a practical application in solving real-world problems.</p><p>Azure ML emerged as a game-changer on this journey. Its user-friendly interface and a wide array of components made complex tasks manageable. The platform’s ease in creating and deploying models, along with its robustness, left a lasting impression on me.</p><p>To those who are planning to take this exam, I can say It’s worth it. The experience is transformative, offering insights and skills that transcend the exam room. Good luck with your exam, may it be as rewarding for you as it was for me!</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=0f655fab49e2" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Microsoft Fabric: all-in-one Enterprise Analytics Platform]]></title>
            <link>https://medium.com/@Ishani.Udeshika/microsoft-fabric-all-in-one-enterprise-analytics-platform-909b10e91c64?source=rss-ead4128890d1------2</link>
            <guid isPermaLink="false">https://medium.com/p/909b10e91c64</guid>
            <category><![CDATA[onelake]]></category>
            <category><![CDATA[enterprise-analytics]]></category>
            <category><![CDATA[microsoft-fabric]]></category>
            <category><![CDATA[power-bi]]></category>
            <category><![CDATA[data-warehouse]]></category>
            <dc:creator><![CDATA[Ishani Udeshika]]></dc:creator>
            <pubDate>Sun, 20 Aug 2023 16:23:34 GMT</pubDate>
            <atom:updated>2023-08-20T16:23:34.892Z</atom:updated>
            <content:encoded><![CDATA[<p>In the current digital era, we are continuously surrounded by a vast amount of data. We are now able to delve deeply into these vast data reservoirs thanks to the power of Data Engineering and analytics, revealing crucial insights that direct our industry’s leaders. Our reliance on data analytics becomes progressively more important as AI advances continue. The amount of data coming in appears to increase every day. Due to this, we require platforms that not only manage these enormous amounts efficiently but also encourage AI’s continuous learning. Platforms for data processing are now at the cutting edge of technology as a result. In May 2023, Microsoft unveiled its comprehensive data engineering and analytics SaaS service, called Microsoft Fabric, in this dynamic market.</p><h4>Why Microsoft Fabric?</h4><p><strong><em>Hundreds of platforms beckon our attention in today’s fast-paced digital landscape. Amidst this deluge, What makes Microsoft Fabric stands out?</em></strong></p><p>Imagine a world where data doesn’t just sit idle; it speaks, guides, it reveals. With the sophisticated algorithms and AI capabilities of Microsoft Fabric, data transforms into a goldmine of actionable intelligence. As our reliance on real-time data insights intensifies, the tools we use must not only deliver but anticipate our next move.</p><p>We’re all accustomed to a digital environment where the ‘now’ is already too late. Platforms need to be more than user-friendly; they need to feel like an extension of our own thought processes. Microsoft Fabric not only fits this mold but carves a unique niche for itself, building bridges where gaps existed and fostering a spirit of effortless collaboration.</p><p>For the data scientists, data engineers, business owners, consider a platform where resources aren’t just abundant but also effortlessly within reach. The agility to adapt, mold, and reintegrate resources without a hitch is a dream Microsoft Fabric brings to fruition.</p><p>Every second, a vast expanse of data emerges. Storage? That’s elementary. With Microsoft Fabric, it’s about integration and interaction. The data remains in a singular space yet remains malleable to our choice of analytics tools, marking a departure from the rigid to the dynamic.</p><p>As the digital tapestry grows more intricate, centralized oversight transforms from being a mere convenience to a critical linchpin. This is where Microsoft Fabric shines, ushering in a streamlined approach to governance, ensuring our endeavors are efficient, compliant, and on the mark.</p><h4>What Microsoft Fabric offer?</h4><p>Microsoft Fabric provides unified end-to-end all-in-one analytics experience by deeply integrated comprehensive suite of data products which starts from data movements to data science, real-time Analytics and business intelligence. Fabric brings new and existing data services from Power BI, Azure Synapse, and Azure Data Factory into a single integrated environment.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*jsP4NpIsWet0l1_7" /></figure><p><strong>Data Engineering Service</strong>: Offers world class Spark data processing platform with great development experiences, enabling data engineers to perform large scale data transformation and share the data through the lakehouse.</p><p><strong>Data Factory</strong>: Combines the simplicity of Power Query features that enable citizen developers to perform data integration seamlessly.</p><p><strong>Data Science</strong>: This enables you to build, deploy and operationalize Machine Learning models seamlessly within fabric eco-system</p><p><strong>Data Warehouse</strong>: Provides enterprise grade SQL Warehousing at scale. It fully separates computing from storage, enabling independent scaling.</p><p><strong>Real-time Analytics</strong>: Unlock the possibilities of analyzing semi-structured and unstructured data generated from various sources such as Apps, IoT devices, human interactions, etc. Real-time analytics is the best in class engine for these types of workloads.</p><p><strong>Power BI: </strong>World’s leading Business Intelligence platform. This integration ensure that business owners can access organizational data estate quickly and intuitively to make better decisions.</p><p><strong>OneLake</strong>: This is the foundation on which all the fabric services are built. This provides unified location to store all the organizational data.</p><p>Above all, Microsoft Fabric provides centralized administration and governance across all the above experiences.</p><p>…</p><h3>References:</h3><p>Microsoft Fabric official page:</p><p><a href="https://www.microsoft.com/en-us/microsoft-fabric">https://www.microsoft.com/en-us/microsoft-fabric</a></p><p>Learn more about Microsoft Fabric:</p><p><a href="https://learn.microsoft.com/en-us/fabric/get-started/microsoft-fabric-overview">https://learn.microsoft.com/en-us/fabric/get-started/microsoft-fabric-overview</a></p><p>Data Analytics for the era of AI:</p><p><a href="https://azure.microsoft.com/en-us/blog/introducing-microsoft-fabric-data-analytics-for-the-era-of-ai/">https://azure.microsoft.com/en-us/blog/introducing-microsoft-fabric-data-analytics-for-the-era-of-ai/</a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=909b10e91c64" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>