Protéger les informations personnelles et l’intégrité du consommateur est devenu une priorité forte en Europe. Selon la nouvelle réglementation générale de protection des données, chaque consommateur et citoyen a le droit de connaître l’usage qui est fait de ses données, et a également le droit de faire supprimer à sa demande toute information le concernant. Les entreprises évoluant sur le marché de l’Union Européenne doivent se conformer à cette nouvelle directive européenne d’ici le 25 Mai 2018.
In the general Business Intelligence field, Power BI is more and more known.
Why choosing Power BI?
A complete decision-making cycle
Power BI is a tool for interactive data visualization in “Self-Service” BI.It is made of 2 separate entities: Power BI Desktop and Power BI Services.
How does it make the difference? Thanks to its 2 entities, Power BI gathers all the stages of Business Intelligence in a one and only tool:
- Multi sources connection
- Data Cleansing and Data Transformation
- Data modeling
- Visualization (Interactive Reports and Dashboards)
Power BI has combined all the Excel’s adds-in* (such as Power Query, Power Pivot, Power view, Power map, …) and has merged them in one unique solution that is rich of functionalities and easy to use at the same time.
Indeed, Power BI offers a wide range of connectors. Thus, it is possible to analyze data from different sources:
- Files ( Excel, Text, XML, …)
- Data bases (Access, SQL Server, …)
- Azure (Data Bases SQL Microsoft Azure, …)
- Online Services (Facebook, Google Analytics, Dynamics 365, Mailchimp, …)
(More information regarding connection à https://powerbi.microsoft.com/en-us/documentation/powerbi-desktop-data-sources/ )
Some integrations to consider
Power BI perfectly integrates perfectly with all the other tools of the Microsoft Office Suite, particularly with Dynamics 365(1) and Sharepoint (2). This perfect matches, enable to simplify access to companies’ reports by centralizing all the data related to the company from Dynamics 365 and Sharepoint to Power BI.
What are the functional advantages concealed in this high potential tool?
Power BI permits to analyze and visualize data in a simple way. It is thus possible for managers with low-level of technical skills to create interactive and personalized reports, to design tailor-made dashboards, all of it in few clicks. Firstly, the user has to pick a type of visualization (map, diagram, …) among a wide range. The choice of visuals is constantly enriched by the community of Power BI users. In fact, they put their own visual creations at disposal for free. (check à https://app.powerbi.com/visuals/ ).Then, the user just has to drag and drop the relevant data on the chosen interactive visual. Hence it is simple to adapt quickly to this tool thanks to basic Excel competencies. Indeed, the reflex acquired using Excel will help getting used to the different features of the solution.
Moreover, Power BI is a user-friendly tool that has been conceived for the use of a maximum of people. In other words, any collaborator in a company may be able to access his data, create reports and share them with their co-workers without the need of technical intermediaries. It is possible to ask questions (such as ratios, formulas, indicators, …) in natural language directly to Power BI, the tool will be able to generate automatic responses without using computer languages. As a matter of fact, the tool is suitable for the use of Cortana (3), the users can directly ask their questions to Cortana and instantly obtain an answer. However, to perform specific analysis, queries are made in DAX (4) language.
A collaborative tool
Power BI provides the possibility to managers to collaborate around interactive reports in the Cloud. Managers can export their reports from Power BI Desktop to Power BI Services or to Azure (5).
Reports can be shared with control over the access rights (consultation right/modification right) to lower risks of data breach. It is even possible to work over the same report at the same time and to consult a Dashboard from different devices.
To complete this collaborative aspect, there is also a free mobile application which permits to easily access reports and dashboards during business trips for example.
An advantageous pricing
This solution is one of the cheapest data visualization tool in the market.
- Power BI Desktop : FREE
- Power BI Service: FREE version of Power BI Services but with volumetry constraints. Pro version for 9.99$ € per user, per month. The choice depends on the size of the company. It is not a problem to work with 2 different types of licence inside the same company. Thus, there will not be any compatible problem for visualizing reports. (want to know more à https://powerbi.microsoft.com/en-us/pricing/ ).
- Power BI Mobile: FREE (available on Google Play, App Store and Windows Store)
What next in term of optimization ?
Microsoft managed to create a well federated community that is reactive to the updates and with a lot of ideas for the improvement of the solution. Thanks to the reactivity of the community, different axes of thinking come up naturally from the Power BI community forum (check http://community.powerbi.com/t5/Forums/ct-p/Forums ). Here are some recommendations raised by the community, that Microsoft took in consideration: It would be interesting to have the possibility to export reports directly on PowerPoint conserving the interactivity of the visuals. Moreover, the ergonomics of the app on I phone is not at its maximum. It may be possible to work on optimizing the visualizations for a better user experience on smartphone.
Nevertheless, a major limit seems to stand out: There is no On-premise version of Power BI for now on the market. However, it seems that an On-premise version could be existing by the end of 2017.
With all these different features, Power BI can still be qualified as “Data Visualization tool” but it differentiates from its peers essentially thanks to its collaborative aspect and to the intuitiveness of use. Power BI reached the top and deserves its position among the leaders of the Self-service BI solutions.
Written by Cléo Beruben, Business Intelligence Consultant at AEROW Decision
Developed by the firm Dataiku, Data Science Studio (DSS) is a tool making easier the creation and deployment of business data-oriented predictive applications. Combining tools for data handling, statistics, visualization and predictive analysis, it allows fastening all necessary steps: plug-in of data, cleaning and data preparation, benchmarking and test of various predictive algorithms.
By featuring visual recipes and analysis for data preparation and machine learning, it is meant to be used by Data Scientists as well as non-developers. The tool is particularly suitable for collaborative work which ensures individual participation in the success of projects and includes connections to the main and most promising data processing and analysis technologies.
Created in 2013, the firm Dataiku has known an ever growing interest from the industry, with successful predictive and prescriptive projects. Today, the firm has customers such as: BlaBlaCar, Vente-privées, Chronopost, Page Jaunes, La Poste, and a few big clients in the U.S.
Connections to most of the data sources of the market
DSS enables direct connections to the most common data sources (Oracle SQL, Microsoft SQL Server, Hadoop, Cassandra, MongoDB, Elasticsearch, S3, etc.) and formats (CSV, Excel, JSON, SAS, Avro, etc.). It also integrates a connection to Spark, an engine dedicated to high performance distributed processing, potentially much faster for handling large data volume than the “MapReduce” programming model of Hadoop.
Data preparation made quickly and easily
Once the data is retrieved, the first step is to clean and prepare it. This process typically takes as much as 80% of the total project duration. To fasten this process, DSS automatically infers data types (such as gender, country, IP address, URL, date, etc.), uses this inference to check the validity of data, and suggests contextual transformations (like getting the country out the IP, the day of week out of the date, etc.). One has instant visual feedback of any operation, and specific statistics views for each data column. Furthermore, DSS handles common tasks such as replacement and imputation of field values, grouping or splitting datasets, joining and aggregating data…; In a visual and intuitive manner to easily clean, format and enrich the data.
The power of machine learning in a few click
DSS is especially made for benchmarking machine learning algorithms by providing an unified platform for data transformation, visualization and modelling. It automatically creates a features engineering pipeline including missing value imputation, dummy variable encoding, feature generation, and feature reduction through PCA. Once data is cleaned, enriched and formatted, the application allows training several supervised or unsupervised models in parallel, using for example algorithms from the popular Scikit-Learn and Spark MLlib (featuring high scalability to large data volume). The training phase can be made using fully integrated cross-validation techniques, or even on-the-fly in case of evolving datasets. One can then visualize and easily compare their performance and tune their parameters in a graphical manner. Apart from the natively available classification and regression algorithms, DSS also interfaces with custom models (as long as they have a Scikit-Learn compatible API). Finally, a history of trained models can be viewed for use in your workflows to easily score new records.
A unified and flexible Interface to deal with every parts of the project, from raw data to predictions
DSS provides a unified graphical interface for data transformation, visualization and modelling. This allows for instance transforming data and visualizing results from predictive models in the same environment, enabling fast iteration cycles, without the need for custom code. If desired, one can also mix these visual recipes from the graphical interface and custom codes written in various programming languages (SQL, Python, R, Scala) or dialects (Pig, Hive, etc.). A synthesized view of the whole workflow (datasets and transformations) can be seen and a smart reconstruction engine can be used to rerun each computation steps and necessary datasets. The processing chain can be scheduled or triggered based on customable conditions (for instance when source data changes)
A collaborative platform enabling agile developments
For the sake of collaborative work, DSS is designed to ease solving conflicts among contributors and can be interfaced with a LDAP repository to deal with user rights. The project management part of DSS has also been made easier via checklists. These organizational aspects and the possibility for fast iteration cycles are key ingredients for agile developments.
A free access to most of the functionalities to have a hand-on expertise and several levels of usage
A free community edition is available and can be found here : http://www.dataiku.com/products/trynow/, offering a large set of the aforementioned functionalities, which allows several levels of usage (data analyst, data scientist, developers and non-developers, etc.).
Written by Selim Seddiki,
Consultant in Business Intelligence engineering – AEROW Decision
L’OpenText Innovation Tour 2017, c’est plus de 2000 participants, 190 sessions et surtout l’occasion d’assister à des sessions pertinentes et spécialisées autour des solutions OpenText et partenaires… Rencontrez nos experts xECM, CSP, VIM et AnswerModules en démo et rejoignez-nous sur ce rendez-vous unique…