Forecast using Neural Network by MAQ Software - Power BI Visual Introduction

00:06:24
https://www.youtube.com/watch?v=UfQR3SHs81o

摘要

TLDRIn this video, Manuel Quintana from Pragmatic Works and MAQ Software introduces a custom Power BI visual known as 'Forecast using Neural Network.' This visual is geared towards predictive analytics by leveraging neural networks, which excel at analyzing nonlinear data patterns. The visual requires the installation of specific R packages, which can be easily done through the Microsoft Store. Once installed, it allows users to input data, such as the historical price of gold, and observe and predict trends over time. The visual provides customization options for colors, axes, and forecast settings. Users can manually adjust advanced settings like the learning decay rate and number of iterations to optimize predictions. The video highlights the visual's ability to offer predictive insights, including the use of confidence intervals to display prediction reliability. For further queries, MAQ Software and Pragmatic Works offer support and training respectively.

心得

  • 🔍 Introducing Forecast using Neural Network, a custom visual for Power BI.
  • 🛠 Installation requires R packages, easily available via Microsoft Store.
  • 📈 Analyze and predict trends using the neural network for nonlinear data.
  • 🎨 Customizable visual settings, including color and forecast parameters.
  • ⏳ Confidence intervals to denote prediction reliability are available.
  • ⚙️ Advanced users can manually adjust forecast settings if desired.
  • ⌛ Recognize that neural network analysis takes time for accurate predictions.
  • 📧 Contact MAQ Software for solutions: sales@maqsoftware.com.
  • 📚 Pragmatic Works offers training for using Power BI: training@pragmaticworks.com.
  • 🌐 Enables use of predictive analytics within Power BI for business insights.

时间轴

  • 00:00:00 - 00:06:24

    Manuel Quintana introduces a video collaboration between Pragmatic Works and MAQ Software focusing on the "Forecast using Neural Network" custom visual in Power BI. This advanced visual leverages data science and predictive analytics. While prerequisites are necessary, the setup is straightforward via the Microsoft Store. The visual uses R packages to analyze data, and its strength lies in interpreting nonlinear data using neural networks. The neural networks help generate predictions based on past observed values, making it a useful tool for data analysts dealing with complex data patterns.

思维导图

视频问答

  • What software does the video feature?

    The video features the 'Forecast using Neural Network' custom visual for Power BI by MAQ Software.

  • Do I need any prerequisites before using this custom visual?

    Yes, you will need to download and install necessary R packages, which are prompted during the visual installation process from the Microsoft Store.

  • What data type is showcased in the video for analysis?

    The showcased data is the price of gold over time.

  • Can I customize the visual settings?

    Yes, you can customize settings such as background color, forecast color, observed color, and more.

  • What is the advantage of using a neural network for predictions in this visual?

    Neural networks are excellent for deriving patterns in nonlinear data.

  • What does the confidence interval in the visual represent?

    The confidence interval shows the range where predicted values are expected to fall, representing prediction reliability.

  • Can I adjust forecast settings manually?

    Yes, you can switch from auto to user-defined settings to manually control parameters like decay, iterations, and units.

  • Who should I contact for troubleshooting or business solutions related to this visual?

    For solutions or questions, contact MAQ Software at sales@maqsoftware.com.

  • Is training offered for using Power BI with this tool?

    Yes, training is offered by Pragmatic Works, and you can reach out via training@pragmaticworks.com.

  • What is the main use of the neural network in this visual?

    It is used for predictive analytics, enabling users to predict future data patterns based on historical data.

查看更多视频摘要

即时访问由人工智能支持的免费 YouTube 视频摘要！
字幕
en
自动滚动:
  • 00:00:06
    Hello and welcome, my name is Manuel Quintana with Pragmatic Works, and in
  • 00:00:10
    collaboration with MAQ Software, we're bringing you today's video to look at
  • 00:00:15
    their custom visual known as Forecast using Neural Network. Now, this is a
  • 00:00:20
    fantastic visual, and it starts reaching into the realms of data science and
  • 00:00:24
    predictive analytics—which is fantastic—but it is a more advanced
  • 00:00:30
    area of conversation. We are gonna see how easy it is to implement this
  • 00:00:35
    custom visual right into your Power BI reports and start creating some
  • 00:00:39
    predictive analytic visuals, which is fantastic. Now, it should be noted that as
  • 00:00:44
    part of this visual you will need to download and install some prerequisites,
  • 00:00:48
    but all of this is done easily for you when you go to the Microsoft Store and
  • 00:00:52
    you go to install this specific visual. You'll be prompted with a message
  • 00:00:56
    letting you know that you do need to have some items installed, and there is a
  • 00:00:59
    very nice conveniently located install button right there, and it will go
  • 00:01:04
    through that process. What we're doing is we're installing the necessary R
  • 00:01:07
    packages for this visual to work. Once that's set and in place, now you can go
  • 00:01:12
    ahead and start using this custom visual and looking at data over time, maybe a
  • 00:01:17
    different type of data series, and then as well as seeing the data that you have
  • 00:01:21
    and you are presenting, we can now predict and get some results back from
  • 00:01:26
    what the neural network algorithm has learned. That is what this is all
  • 00:01:30
    about: leveraging the neural network learning algorithm, which is also known
  • 00:01:35
    like as a black box algorithm or a deep learning algorithm, and it's really great
  • 00:01:40
    at looking at nonlinear data. There are quite a few visuals out in the
  • 00:01:44
    marketplace that do predictive analytics—and they're more focused around using
  • 00:01:48
    algorithms that are great at handling linear data—but where neural network
  • 00:01:52
    shines is deriving patterns where it is nonlinear, which can be rather difficult.
  • 00:01:56
    Hopefully you're excited and you're ready to enjoy. Let's head over to Power
  • 00:02:00
    BI and see how we can put this custom visual into play.
  • 00:02:04
    Here we are looking at the Forecast using Neural Network from MAQ Software.
  • 00:02:09
    The nature of our data here is just looking at, if we look at the data view,
  • 00:02:14
    just the price of gold over time—so over years. Very straightforward, and that's
  • 00:02:18
    what we've input. If we look at the field well here for our custom visual, the
  • 00:02:23
    series can either be a time or numeric series and then our value. Now do note
  • 00:02:28
    that what we're seeing in this teal-ish color—that is our observed values. The
  • 00:02:32
    yellow is was what our predicted values are. So to look through this, lets go
  • 00:02:37
    right into the format area, so we can have an understanding of where we can
  • 00:02:40
    control this. The plot setting is where we can dictate the background color, the
  • 00:02:44
    forecast color, and the observed color, which is very important. Of course, we can
  • 00:02:49
    control the x and the y axis, but the other main element here is going to be
  • 00:02:53
    the forecast settings. Now, the default is setting this to auto, but if you have a
  • 00:02:58
    good understanding and you feel confident, you can actually control the
  • 00:03:02
    parameters that drive the neural network. We'll look at this momentarily, but I did
  • 00:03:07
    want to display it. Once you have this configured—and it should be noted—it
  • 00:03:10
    does take some time. Once you define the fields that are going to populate your
  • 00:03:14
    visual, the algorithm—the neural network—needs to run. It's going to be doing
  • 00:03:19
    analysis and creating predictions of your data. Don't worry if it's taking
  • 00:03:23
    a moment; that is the nature of this type of a visual. We can see that we do have
  • 00:03:27
    some really neat capabilities with inside of it where we can actually draw
  • 00:03:30
    boxes and zoom in to certain elements. By double clicking anywhere in the box, you
  • 00:03:34
    go back to your main view. You can enable some spike lines, so as you're moving
  • 00:03:38
    across, you can see how that correlates to the rest of the data—how we can see
  • 00:03:41
    this is a rise and there's no other peak that equals this one as far as looking
  • 00:03:45
    back in time. Of course, you may notice when we zoom in, in this area just
  • 00:03:50
    here that we have this little shaded area. This relates to the confidence
  • 00:03:55
    intervals. By default, this is turned off, but by turning this on, you now get a
  • 00:03:59
    range. You can choose—or you can see we have some confidence levels we can
  • 00:04:03
    choose with—lowering this number: having less confidence is gonna narrow this
  • 00:04:08
    range. The higher the number we're giving ourselves a little more breathing
  • 00:04:12
    room saying hey we're confident that the values in this time frame will fall
  • 00:04:16
    between these little brackets, and as you hover over,
  • 00:04:18
    you can see the information. The confidence level here for 2016 is gonna
  • 00:04:22
    be that 1373 and some change, while the yellow line is just the raw
  • 00:04:29
    predicted line. Really neat, really powerful, but as I mentioned, we could go
  • 00:04:33
    even further. If we look at the same visual but in the context of going over
  • 00:04:38
    to the format area and choosing to switch the forecast settings from auto
  • 00:04:42
    to user-defined, we have a couple of choices here. Now, of course, there's a lot
  • 00:04:47
    that goes into data science and learning, but we have some items here where
  • 00:04:51
    decay kind of controls the learning rate of the neural network. The maximum
  • 00:04:55
    number of iterations is how many times it's gonna run these numbers through the
  • 00:04:58
    neural network and coming up with different values and then coming back
  • 00:05:01
    with the best distribution. The number of units, in this case, is gonna represent
  • 00:05:06
    our series, which were going in years, so this is going to predict out to ten
  • 00:05:10
    years from where we ended our observed. Of course, epochs is going to be the kind
  • 00:05:15
    of rotations, so we're gonna have two hundred iterations over a single epoch.
  • 00:05:19
    Here, we're gonna have eight epochs. Lowering these values, of course, has less
  • 00:05:24
    iterations—there's less loading time—but there is this concept of overfitting,
  • 00:05:29
    which can occur if you start increasing these numbers. Basically, you're making it
  • 00:05:33
    learn the specifics of this data, so when new data is introduced, it may not
  • 00:05:38
    interpret that as correctly. Like I said, we're working in the realm of data
  • 00:05:41
    science, so there's a lot of information to understand here, but hopefully you can
  • 00:05:45
    see very quickly and how easily you can now already start using data science or
  • 00:05:51
    predictive model visuals here right within Power BI with the usage of this
  • 00:05:57
    Forecast using Neural Networks by MAQ Software. Hopefully you enjoyed, and thanks
  • 00:06:02
    for watching our video. If you have any questions about this visual or need a
  • 00:06:06
    similar business solution, feel free to contact MAQ Software at sales@maqsoftware.com.
  • 00:06:13
    As well, for any of your Power BI training needs, be sure to reach
  • 00:06:17
    out to us at Pragmatic Works by emailing training@pragmaticworks.com.
标签
  • Forecast using Neural Network
  • Power BI
  • Predictive Analytics
  • Neural Network
  • MAQ Software
  • Data Science
  • R Packages
  • Customization
  • Confidence Interval
  • Power BI Training