<?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 Osama Bin Yaqoob on Medium]]></title>
        <description><![CDATA[Stories by Osama Bin Yaqoob on Medium]]></description>
        <link>https://medium.com/@osamayaqoob58?source=rss-17c0588769ca------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/0*uLuV3EQSXSJqRrMP</url>
            <title>Stories by Osama Bin Yaqoob on Medium</title>
            <link>https://medium.com/@osamayaqoob58?source=rss-17c0588769ca------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Sat, 21 Nov 2026 21:39:54 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@osamayaqoob58/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[Comparative Analysis of LSTM and FB-Prophet for Time Series Forecasting]]></title>
            <link>https://medium.com/@osamayaqoob58/comparative-analysis-of-lstm-and-fb-prophet-for-time-series-forecasting-cfe14de48fb9?source=rss-17c0588769ca------2</link>
            <guid isPermaLink="false">https://medium.com/p/cfe14de48fb9</guid>
            <dc:creator><![CDATA[Osama Bin Yaqoob]]></dc:creator>
            <pubDate>Tue, 11 Jun 2024 06:11:41 GMT</pubDate>
            <atom:updated>2024-06-11T07:50:00.627Z</atom:updated>
            <content:encoded><![CDATA[<p>Recurrent neural networks (RNNs) with Long Short-Term Memory (LSTM) architecture are better at capturing temporal dependencies and overcoming the vanishing gradient issue. This flowchart and explanation are provided.<br>Flow Diagram for LSTM</p><p><strong>Input Sequence</strong>:</p><ul><li>Input data is sequentially fed into the LSTM network.</li></ul><p><strong>LSTM Cell</strong>:</p><p>Each LSTM cell has three gates: input gate, forget gate, and output gate.</p><ul><li><strong>Forget Gate</strong>: Decides what information to discard from the cell state.</li><li><strong>Input Gate</strong>: Decides which values from the input will update the cell state.</li><li><strong>Output Gate</strong>: Decides what the next hidden state should be based on the cell state.</li></ul><p><strong>Cell State and Hidden State</strong>:</p><ul><li>The cell state carries the memory of the network, allowing it to maintain long-term dependencies.</li><li>The hidden state is updated and passed to the next LSTM cell.</li></ul><p><strong>Output Layer</strong>:</p><p>The final hidden state is used to make predictions</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*GmBw2q8HYEXEDUO8.png" /><figcaption>LSTM Flow Diagram</figcaption></figure><h3>How FB-Prophet Works</h3><p><strong>FB-Prophet</strong> is an open-source forecasting tool developed by Facebook, designed for time series data. It is particularly effective for capturing seasonality, holidays, and trend patterns in data.</p><h4>Flow for FB-Prophet</h4><p><strong>Input Data</strong>:</p><p>Input data consists of a time series with timestamps and corresponding values.</p><p><strong>Pre-processing</strong>:</p><ul><li>Handles missing data and outliers.</li><li>Allows for holidays and special events to be included.</li></ul><p><strong>Model Components</strong>:</p><ul><li><strong>Trend</strong>: Captures the overall trend of the time series.</li><li><strong>Seasonality</strong>: Captures weekly, monthly, or yearly seasonality patterns.</li><li><strong>Holidays</strong>: Captures the effect of holidays and special events.</li></ul><p><strong>Fitting the Model</strong>:</p><p>The model fits each component to the data separately.</p><p><strong>Prediction</strong>:</p><ul><li>Combines the components to make predictions.</li><li>Provides uncertainty intervals for the forecasts.</li></ul><h3>Advantages and Disadvantages</h3><h4>LSTM</h4><p><strong>Advantages</strong>:</p><ol><li><strong>Captures Long-Term Dependencies</strong>: LSTMs can capture dependencies in data over long periods.</li><li><strong>Effective with Sequential Data</strong>: Suitable for tasks like time series prediction, speech recognition, and text generation.</li><li><strong>Flexible Architecture</strong>: Can be stacked to form deep networks, enhancing model capacity.</li></ol><p><strong>Disadvantages</strong>:</p><ol><li><strong>Complex and Computationally Expensive</strong>: LSTMs are more complex and require more computational power and time to train.</li><li><strong>Sensitive to Hyperparameters</strong>: Requires careful tuning of hyperparameters.</li><li><strong>Difficult to Interpret</strong>: The internal workings of LSTM models are often seen as black boxes, making them harder to interpret.</li></ol><h4>FB-Prophet</h4><p><strong>Advantages</strong>:</p><ol><li><strong>User-Friendly</strong>: Easy to use with simple parameter tuning, making it accessible to non-experts.</li><li><strong>Handles Missing Data</strong>: Robust to missing data and outliers.</li><li><strong>Captures Seasonality and Holidays</strong>: Effectively models seasonality and special events without extensive tuning.</li></ol><p><strong>Disadvantages</strong>:</p><ol><li><strong>Limited to Additive Models</strong>: Assumes that the effects of trend, seasonality, and holidays are additive, which may not suit all time series data.</li><li><strong>Less Flexible with Complex Patterns</strong>: May not capture complex patterns as effectively as more advanced models like LSTM.</li><li><strong>Limited Customization</strong>: While easy to use, it offers limited customization for advanced users compared to deep learning models.</li></ol><h3>Conclusion</h3><p>Both LSTM and FB-Prophet are powerful tools for time series forecasting, each with its strengths and weaknesses. LSTM is highly effective for capturing complex temporal dependencies but requires more expertise and computational resources. FB-Prophet, on the other hand, is user-friendly and effective for capturing seasonality and trends, making it ideal for straightforward time series forecasting tasks. The choice between the two depends on the specific requirements of the task and the expertise available.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=cfe14de48fb9" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>