Please describe your organization.
Neal Analytics is a predictive analytics consulting firm. We work with customers that target use cases which will drive value and return on investment [ROI] for their business. We typically engage on an executive level, proving ROI for a particular use case. We refine said use case and deliver a predictive algorithm into production in order to help improve a particular operational process, feature, or aspect of a business. We engage in business with large enterprise companies that are interested in improving a process to drive a tangible result. Typically, these firms operate in manufacturing, retail, consumer goods, energy, oil and gas, raw materials, and so on. We also work with education companies and other types of businesses. A number of our clients are also technology companies.
We also offer traditional business intelligence [BI] and data warehouse implementation. We run the gamut when it comes to dealing with data, but we only take on businesses that align with our strategic focus, namely helping customers increase their ROI on data assets.
Right now, we employ 30-40 data scientists and engineers across the company. We have an office in Seattle and one in India, where our resources are located. We have been one of the leading implementers of Azure Machine Learning, Microsoft's predictive analytics product. We typically implement technology through their SQL server product for data warehousing and use HDInsight for big data. We use Azure data factory for data movement and a host of other Azure-based technologies.
The use cases that we shoot for within predictive analytics include demand forecasting, decision modeling, resource prediction, predictive maintenance, systems integration work, and other similar cases. We focus on the use cases where we can drive the largest amount of ROI for data.
What is your position?
I am currently transitioning to the role of alliances manager.
Please discuss your experience with PowerBI.
PowerBI is our visualization tool of choice. Often times, we use its features in order to do our data profiling and visualization. It's part of our data science and business intelligence workflows, because visualization is the end-product in these cases. It's also part of our data warehousing workflow. We use PowerBI extensively.
We gained a competency a couple of years ago in Azure Machine Learning, when the product was released. Since then, we've gained a competency in PowerBI. I lead a local group focused on the solution. Since that group has taken off, we've gained competency particularly around content packs and other features.
Predictive analytics is, at the same time, different from business intelligence, and similar to it. Ultimately, the goal is to surface predictions within an existing business intelligence system. In order to make such predictions, a completely different set of technologies will be used. A client can still do what they’re doing with complex SQL queries and hardcore tabular logic. Predictive logic can be achieved in both places, the difference being how long it would take and how good the results are going to be. It's also important to distinguish between data management and predictive analytics. People will often consider HDInsight to be predictive analytics when, in fact, it's the place where data is being dumped. Predictive analytics are being done in SparkR, on top of HDInsight. It's important to distinguish between the big data platform and the predictive analytics toolkit on top, and whatever business intelligence toolkit a client is using. Some of the available toolkits, like Click, use Hadoop as a backend.
Could you provide a walkthrough for a typical case study and discuss the software in greater detail?
We've recently wrapped up a project with a local municipal school district which is getting press write-ups because of the efficacy of PowerBI. I presented a case study at Microsoft's Global Education Partners Summit.
PowerBI spans between Office365 and Azure because it's a business intelligence workload. The data-heavy assets fall into the Azure category, while most of the per-user assets that Microsoft has fall into the Office365 category. As such, PowerBI has an important placement among Microsoft's offerings.
Overall, the major strength of PowerBI is its licensing structure. It's designed to fit in with Office365, which is very convenient for larger enterprises that want to add BI capabilities and give their users the functionalities that they require. That is a decidedly different model to the one offered by Tableau. They typically charge $500,000 for server-side licensing, the internal IT organization being required to provide the server. PowerBI is essentially a free managed service on Azure, which Microsoft offers in addition to their license. This means that a client is also paying for server management. It's important to factor this in when considering business intelligence technologies: where and how is the data going to be hosted, and how does that model fit into general practices?
One of the major strengths that PowerBI offers are content packs. They're an essential connector, similar to what Tableau offers. The difference is that they enable publishing environmental sets of pre-canned visualizations off of a configurable connector. This means that people who want to pull Google Analytics data, can type in their Google account information and get a visualization right away. This increases the time value that PowerBI offers as a business intelligence suite. It means that information can be ascribed to one location and visualizations can be made rapidly.
PowerBI includes all the power tools that Microsoft has been bundling and testing at a grand scale: Power Query, Power Pivot, Power VU and Power Map. Power Map isn't fully integrated into PowerBI yet, but the other ones are. They are important features that literally no one else has.
Microsoft acquired a company called FoxPro some time ago, which offered them Rushmore Indexing Technology. It equated to a database index being done in-memory. Since then, in-memory computing has become significant when dealing with data. With the advent of solid-state drives, it has become more feasible to access data rapidly. It's the reason for Amazon Redshift's success. It's backed by the right hardware in order to access data quickly. Their technology isn't particularly great, but they have the right hardware behind it.
PowerBI uses a historical Microsoft technology that was acquired and developed continuously and which uses its database engine for in-memory, real-time processing. When booting an integrated development environment for PowerBI, a user is developing on the computer's memory. All the data that is pulled in, stays within the computer's memory. When that same data is uploaded to a PowerBI server, it will continue to live in that server's memory, which is very useful for rapid interactive visualization and quick development. PowerBI lets users get into the data and alter its structure. I can have a semantic modeling process within the tool. No other business intelligence tool will let me do that as seamlessly as PowerBI.
On MSBIacademy.com, there are a couple of videos regarding BI semantic modeling. They explain more in-depth how different technologies stack pieces within the Microsoft PowerBI ecosystem. They offer the full story, going from Excel, to BI, to SQL server analysis, to a services cube, all published from PowerBI. With SQL Server 2016, they're including the new query languages that they've introduced through the PowerBI toolset: DAX, M and R.
I can save an ETL built in Power Query and have it run on my SQL server. It scales all the way up to the enterprise-level technology offered by Microsoft, which is an extremely compelling story when it comes to business intelligence. Not only does it scale across users seamlessly, without having to manage server-side infrastructure, but it scales across the platform that Microsoft has when it comes to technology. It lets you do development and scale it as your organization grows.
Are there any areas that could be added to or improved upon when it comes to PowerBI technology?
The biggest strength of PowerBI is also its biggest weakness: it's on a rapid development cycle, so things are always changing. Unless a user documents themselves on new technology every other week, they're going to get lost. Microsoft has a monthly or bi-monthly release, with new bits going into the product constantly. This is useful for professionals that deal with the product every day, but it's bad for casual users. They won't read the patch notes and won't have an understanding of what's going on.
Tableau lets users visualize bad data. PowerBI forces them to get their data right. Tableau doesn't offer the ability to correct data, while PowerBI does. Tableau is an enterprise warehouse salesman's best friend because it lets people visualize simply what's available. If your visualizations are bad, you should pay $600 for a big data warehouse deployment project and create some cubes with the perspective that you want, in order to do your visualization. PowerBI will let users clean data and do the first step of building cubes and the semantic model that they need for doing the express visualization themselves. It can then be scaled up within SQL Server Analysis Services in PowerBI, or from PowerBI straight to that technology layer, and escalated within the technology stack. This is extremely valuable because Microsoft is a full platform company, while Tableau is a one-off solution. When it comes to other visualization platforms, I've not really seen Oracle or IBM compete with PowerBI. They do offer solutions, but whenever I'm called in by a client, it's for the purpose of ripping and replacing Oracle or IBM solutions with those from Microsoft. The price point is lower and the product suite is better. The only real competitor to Microsoft is Tableau, who are also a user of the Microsoft business intelligence stack.
Have you ever used Microsoft's support team for PowerBI or referenced their support resources? If so, how would you categorize that experience?
I know most of their product team because the meet-up group that I lead gets together in the Microsoft buildings. If I have an issue, I send emails to the people that built the product, who engage with me in order to figure it out. My support experience has been stellar, and I can't emphasize enough how great they are, but I have not gone through support through normal channels.
I would advise any customer that is considering PowerBI to find the appropriate partner for what they're trying to do within the Microsoft ecosystem, and to leverage their help. Doing this will lead to having deeper contacts than any other client.
We have five additional questions. For each of these, we ask that you rate PowerBI on a scale of one to five, with five being the best score.
How would you rate PowerBI for its available features?
Four and a half.
How would you rate it for ease of use?
Five. However this depends on the user. PowerBI is great for someone with a background in business intelligence semantic modeling and database theory. It was built for the management information systems undergrad. It's the perfect toolkit for a person with this type of education to use in order to understand how things work. If the user has ever worked in Access, they'll love PowerBI. It is what Access was meant to be, without all the useless forms. However, it's not going to be the most intuitive environment for a data scientist. It has a different modeling paradigm that what they're used to. Instead of doing econometrics statistical modeling, they'll have to do tabular modeling and build out semantic environments. If someone has a PhD in statistics or a background in econometrics or biology, their enjoyment of PowerBI will be lower. The training they received in school subscribes to a different paradigm.
How likely are you to recommend PowerBI to a friend or colleague, out of five?
Five. I believe I mentioned that I lead the local meet-up group. I recommend it for free and without solicitation.
What would be your overall rating for PowerBI?